Configuration Reference#

The JupyterHub Helm chart is configurable by values in your config.yaml. In this way, you can extend user resources, build off of different Docker images, manage security and authentication, and more.

Below is a description of many but not all of the configurable values for the Helm chart. To see all configurable options, inspect their default values defined here.

For more guided information about some specific things you can do with modifications to the helm chart, see the Customization Guide.

enabled#

enabled is ignored by the jupyterhub chart itself, but a chart depending on the jupyterhub chart conditionally can make use this config option as the condition.

fullnameOverride#

fullnameOverride and nameOverride allow you to adjust how the resources part of the Helm chart are named.

Name format

Resource types

fullnameOverride

nameOverride

Note

component

namespaced

""

*

Default

release-component

cluster wide

""

*

Default

fullname-component

*

str

*

-

release-component

*

null

""

-

release-(name-)component

*

null

str

omitted if contained in release

release-(chart-)component

*

null

null

omitted if contained in release

Warning!

Changing fullnameOverride or nameOverride after the initial installation of the chart isn’t supported. Changing their values likely leads to a reset of non-external JupyterHub databases, abandonment of users’ storage, and severed couplings to currently running user pods.

If you are a developer of a chart depending on this chart, you should avoid hardcoding names. If you want to reference the name of a resource in this chart from a parent helm chart’s template, you can make use of the global named templates instead.

# some pod definition of a parent chart helm template
schedulerName: {{ include "jupyterhub.user-scheduler.fullname" . }}

To access them from a container, you can also rely on the hub ConfigMap that contains entries of all the resource names.

# some container definition in a parent chart helm template
env:
  - name: SCHEDULER_NAME
    valueFrom:
      configMapKeyRef:
        name: {{ include "jupyterhub.user-scheduler.fullname" . }}
        key: user-scheduler

nameOverride#

See the documentation under fullnameOverride.

imagePullSecret#

imagePullSecret.create#

Toggle the creation of the k8s Secret with provided credentials to access a private image registry.

Default: false

imagePullSecret.automaticReferenceInjection#

Toggle the automatic reference injection of the created Secret to all pods’ spec.imagePullSecrets configuration.

Default: true

imagePullSecret.registry#

Name of the private registry you want to create a credential set for. It will default to Docker Hub’s image registry.

Examples:

  • https://index.docker.io/v1/

  • quay.io

  • eu.gcr.io

  • alexmorreale.privatereg.net

imagePullSecret.username#

Name of the user you want to use to connect to your private registry.

For external gcr.io, you will use the _json_key.

Examples:

  • alexmorreale

  • alex@pfc.com

  • _json_key

imagePullSecret.password#

Password for the private image registry’s user.

Examples:

  • plaintextpassword

  • abc123SECRETzyx098

For gcr.io registries the password will be a big JSON blob for a Google cloud service account, it should look something like below.

password: |-
  {
    "type": "service_account",
    "project_id": "jupyter-se",
    "private_key_id": "f2ba09118a8d3123b3321bd9a7d6d0d9dc6fdb85",
    ...
  }

imagePullSecret.email#

Specification of an email is most often not required, but it is supported.

imagePullSecrets#

Chart wide configuration to append k8s Secret references to all its pod’s spec.imagePullSecrets configuration.

This will not override or get overridden by pod specific configuration, but instead augment the pod specific configuration.

You can use both the k8s native syntax, where each list element is like {"name": "my-secret-name"}, or you can let list elements be strings naming the secrets directly.

Default: []

hub#

hub.revisionHistoryLimit#

Configures the resource’s spec.revisionHistoryLimit. This is available for Deployment, StatefulSet, and DaemonSet resources.

See the Kubernetes docs for more info.

hub.config#

JupyterHub and its components (authenticators, spawners, etc), are Python classes that expose its configuration through traitlets. With this Helm chart configuration (hub.config), you can directly configure the Python classes through static YAML values. To dynamically set values, you need to use hub.extraConfig instead.

Currently intended only for auth config

This config currently (0.11.0) only influence the software in the hub Pod, but some Helm chart config options such as hub.baseUrl is used to set JupyterHub.base_url in the hub Pod and influence how other Helm templates are rendered.

As we have not yet mapped out all the potential configuration conflicts except for the authentication related configuration options, please accept that using it for something else at this point can lead to issues.

Example

If you inspect documentation or some jupyterhub_config.py to contain the following section:

c.JupyterHub.admin_access = true
c.JupyterHub.admin_users = ["jovyan1", "jovyan2"]
c.KubeSpawner.k8s_api_request_timeout = 10
c.GitHubOAuthenticator.allowed_organizations = ["jupyterhub"]

Then, you would be able to represent it with this configuration like:

hub:
  config:
    JupyterHub:
      admin_access: true
      admin_users:
        - jovyan1
        - jovyan2
    KubeSpawner:
      k8s_api_request_timeout: 10
    GitHubOAuthenticator:
      allowed_organizations:
        - jupyterhub

YAML limitations

You can’t represent Python Bytes or Set objects in YAML directly.

Helm value merging

helm merges a Helm chart’s default values with values passed with the --values or -f flag. During merging, lists are replaced while dictionaries are updated.

hub.extraFiles#

A dictionary with extra files to be injected into the pod’s container on startup. This can for example be used to inject: configuration files, custom user interface templates, images, and more.

# NOTE: "hub" is used in this example, but the configuration is the
#       same for "singleuser".
hub:
  extraFiles:
    # The file key is just a reference that doesn't influence the
    # actual file name.
    <file key>:
      # mountPath is required and must be the absolute file path.
      mountPath: <full file path>

      # Choose one out of the three ways to represent the actual file
      # content: data, stringData, or binaryData.
      #
      # data should be set to a mapping (dictionary). It will in the
      # end be rendered to either YAML, JSON, or TOML based on the
      # filename extension that are required to be either .yaml, .yml,
      # .json, or .toml.
      #
      # If your content is YAML, JSON, or TOML, it can make sense to
      # use data to represent it over stringData as data can be merged
      # instead of replaced if set partially from separate Helm
      # configuration files.
      #
      # Both stringData and binaryData should be set to a string
      # representing the content, where binaryData should be the
      # base64 encoding of the actual file content.
      #
      data:
        myConfig:
          myMap:
            number: 123
            string: "hi"
          myList:
            - 1
            - 2
      stringData: |
        hello world!
      binaryData: aGVsbG8gd29ybGQhCg==

      # mode is by default 0644 and you can optionally override it
      # either by octal notation (example: 0400) or decimal notation
      # (example: 256).
      mode: <file system permissions>

Using –set-file

To avoid embedding entire files in the Helm chart configuration, you can use the --set-file flag during helm upgrade to set the stringData or binaryData field.

hub:
  extraFiles:
    my_image:
      mountPath: /usr/local/share/jupyterhub/static/my_image.png

    # Files in /usr/local/etc/jupyterhub/jupyterhub_config.d are
    # automatically loaded in alphabetical order of the final file
    # name when JupyterHub starts.
    my_config:
      mountPath: /usr/local/etc/jupyterhub/jupyterhub_config.d/my_jupyterhub_config.py
# --set-file expects a text based file, so you need to base64 encode
# it manually first.
base64 my_image.png > my_image.png.b64

helm upgrade <...> \
    --set-file hub.extraFiles.my_image.binaryData=./my_image.png.b64 \
    --set-file hub.extraFiles.my_config.stringData=./my_jupyterhub_config.py

Common uses

  1. JupyterHub template customization

    You can replace the default JupyterHub user interface templates in the hub pod by injecting new ones to /usr/local/share/jupyterhub/templates. These can in turn reference custom images injected to /usr/local/share/jupyterhub/static.

  2. JupyterHub standalone file config

    Instead of embedding JupyterHub python configuration as a string within a YAML file through hub.extraConfig, you can inject a standalone .py file into /usr/local/etc/jupyterhub/jupyterhub_config.d that is automatically loaded.

  3. Flexible configuration

    By injecting files, you don’t have to embed them in a docker image that you have to rebuild.

    If your configuration file is a YAML/JSON/TOML file, you can also use data instead of stringData which allow you to set various configuration in separate Helm config files. This can be useful to help dependent charts override only some configuration part of the file, or to allow for the configuration be set through multiple Helm configuration files.

Limitations

  1. File size

    The files in hub.extraFiles and singleuser.extraFiles are respectively stored in their own k8s Secret resource. As k8s Secret’s are limited, typically to 1MB, you will be limited to a total file size of less than 1MB as there is also base64 encoding that takes place reducing available capacity to 75%.

  2. File updates

    The files that are mounted are only set during container startup. This is because we use subPath as is required to avoid replacing the content of the entire directory we mount in.

hub.baseUrl#

This is the equivalent of c.JupyterHub.base_url, but it is also needed by the Helm chart in general. So, instead of setting c.JupyterHub.base_url, use this configuration.

Default: "/"

hub.command#

A list of strings to be used to replace the JupyterHub image’s ENTRYPOINT entry. Note that in k8s lingo, the Dockerfile’s ENTRYPOINT is called command. The list of strings will be expanded with Helm’s template function tpl which can render Helm template logic inside curly braces ({{... }}).

This could be useful to wrap the invocation of JupyterHub itself in some custom way.

For more details, see the Kubernetes documentation.

Default: []

hub.args#

A list of strings to be used to replace the JupyterHub image’s CMD entry as well as the Helm chart’s default way to start JupyterHub. Note that in k8s lingo, the Dockerfile’s CMD is called args. The list of strings will be expanded with Helm’s template function tpl which can render Helm template logic inside curly braces ({{... }}).

Warning

By replacing the entire configuration file, which is mounted to /usr/local/etc/jupyterhub/jupyterhub_config.py by the Helm chart, instead of appending to it with hub.extraConfig, you expose your deployment for issues stemming from getting out of sync with the Helm chart’s config file.

These kind of issues will be significantly harder to debug and diagnose, and can due to this could cause a lot of time expenditure for both the community maintaining the Helm chart as well as yourself, even if this wasn’t the reason for the issue.

Due to this, we ask that you do your _absolute best to avoid replacing the default provided jupyterhub_config.py file. It can often be possible. For example, if your goal is to have a dedicated .py file for more extensive additions that you can syntax highlight and such and feel limited by passing code in hub.extraConfig which is part of a YAML file, you can use this trick instead.

hub:
  args:
    - "jupyterhub"
    - "--config"
    - "/usr/local/etc/jupyterhub/jupyterhub_config.py"
    - "--debug"
    - "--upgrade-db"

For more details, see the Kubernetes documentation.

Default: []

hub.cookieSecret#

Note

As of version 1.0.0 this will automatically be generated and there is no need to set it manually.

If you wish to reset a generated key, you can use kubectl edit on the k8s Secret typically named hub and remove the hub.config.JupyterHub.cookie_secret entry in the k8s Secret, then perform a new helm upgrade.

A 32-byte cryptographically secure randomly generated string used to sign values of secure cookies set by the hub. If unset, jupyterhub will generate one on startup and save it in the file jupyterhub_cookie_secret in the /srv/jupyterhub directory of the hub container. A value set here will make JupyterHub overwrite any previous file.

You do not need to set this at all if you are using the default configuration for storing databases - sqlite on a persistent volume (with hub.db.type set to the default sqlite-pvc). If you are using an external database, then you must set this value explicitly - or your users will keep getting logged out each time the hub pod restarts.

Changing this value will all user logins to be invalidated. If this secret leaks, immediately change it to something else, or user data can be compromised

# to generate a value, run
openssl rand -hex 32

hub.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

hub.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "quay.io/jupyterhub/k8s-hub"

hub.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a

hub.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

hub.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

hub.networkPolicy#

This configuration regards the creation and configuration of a k8s NetworkPolicy resource.

hub.networkPolicy.enabled#

Toggle the creation of the NetworkPolicy resource targeting this pod, and by doing so, restricting its communication to only what is explicitly allowed in the NetworkPolicy.

Default: true

hub.networkPolicy.ingress#

Additional ingress rules to add besides those that are required for core functionality.

Default: []

hub.networkPolicy.egress#

Additional egress rules to add besides those that are required for core functionality and those added via .egressAllowRules.

Changed in version 2.0.0: The default value changed from providing one very permissive rule allowing all egress to providing no rule. The permissive rule is still provided via .egressAllowRules set to true though.

As an example, below is a configuration that disables the more broadly permissive .privateIPs egress allow rule for the hub pod, and instead provides tightly scoped permissions to access a specific k8s local service as identified by pod labels.

hub:
  networkPolicy:
    egressAllowRules:
      privateIPs: false
    egress:
      - to:
          - podSelector:
              matchLabels:
                app: my-k8s-local-service
        ports:
          - protocol: TCP
            port: 5978

Default: []

hub.networkPolicy.egressAllowRules#

This is a set of predefined rules that when enabled will be added to the NetworkPolicy list of egress rules.

The resulting egress rules will be a composition of:

  • rules specific for the respective pod(s) function within the Helm chart

  • rules based on enabled egressAllowRules flags

  • rules explicitly specified by the user

Note

Each flag under this configuration will not render into a dedicated rule in the NetworkPolicy resource, but instead combine with the other flags to a reduced set of rules to avoid a performance penalty.

New in version 2.0.0.

hub.networkPolicy.egressAllowRules.cloudMetadataServer#

Defaults to false for singleuser servers, but to true for all other network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the cloud metadata server.

Note that the nonPrivateIPs rule is allowing all non Private IP ranges but makes an exception for the cloud metadata server, leaving this as the definitive configuration to allow access to the cloud metadata server.

Changed in version 3.0.0: This configuration is not allowed to be configured true at the same time as singleuser.cloudMetadata.blockWithIptables to avoid an ambiguous configuration.

Default: true

hub.networkPolicy.egressAllowRules.dnsPortsCloudMetadataServer#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the cloud metadata server via port 53.

Relying on this rule for the singleuser config should go hand in hand with disabling singleuser.cloudMetadata.blockWithIptables to avoid an ambiguous configuration.

Known situations when this rule can be relevant:

  • In GKE clusters with Cloud DNS that is reached at the cloud metadata server’s non-private IP.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

New in version 3.0.0.

Default: true

hub.networkPolicy.egressAllowRules.dnsPortsKubeSystemNamespace#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to pods in the kube-system namespace via port 53.

Known situations when this rule can be relevant:

  • GKE, EKS, AKS, and other clusters relying directly on kube-dns or coredns pods in the kube-system namespace.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

New in version 3.0.0.

Default: true

hub.networkPolicy.egressAllowRules.dnsPortsPrivateIPs#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to private IPs via port 53.

Known situations when this rule can be relevant:

  • GKE clusters relying on a DNS server indirectly via a a node local DNS cache at an unknown private IP.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

Warning

This rule is not expected to work in clusters relying on Cilium to enforce the NetworkPolicy rules (includes GKE clusters with Dataplane v2), this is due to a known limitation.

Default: true

hub.networkPolicy.egressAllowRules.nonPrivateIPs#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the non-private IP ranges with the exception of the cloud metadata server. This means respective pod(s) can establish connections to the internet but not (say) an unsecured prometheus server running in the same cluster.

Default: true

hub.networkPolicy.egressAllowRules.privateIPs#

Defaults to false for singleuser servers, but to true for all other network policies.

Private IPs refer to the IP ranges 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16.

When enabled this rule allows the respective pod(s) to establish outbound connections to the internal k8s cluster. This means users can access the internet but not (say) an unsecured prometheus server running in the same cluster.

Since not all workloads in the k8s cluster may have NetworkPolicies setup to restrict their incoming connections, having this set to false can be a good defense against malicious intent from someone in control of software in these pods.

If possible, try to avoid setting this to true as it gives broad permissions that could be specified more directly via the .egress.

Warning

This rule is not expected to work in clusters relying on Cilium to enforce the NetworkPolicy rules (includes GKE clusters with Dataplane v2), this is due to a known limitation.

Default: true

hub.networkPolicy.interNamespaceAccessLabels#

This configuration option determines if both namespaces and pods in other namespaces, that have specific access labels, should be accepted to allow ingress (set to accept), or, if the labels are to be ignored when applied outside the local namespace (set to ignore).

The available access labels for respective NetworkPolicy resources are:

  • hub.jupyter.org/network-access-hub: "true" (hub)

  • hub.jupyter.org/network-access-proxy-http: "true" (proxy.chp, proxy.traefik)

  • hub.jupyter.org/network-access-proxy-api: "true" (proxy.chp)

  • hub.jupyter.org/network-access-singleuser: "true" (singleuser)

Default: "ignore"

hub.networkPolicy.allowedIngressPorts#

A rule to allow ingress on these ports will be added no matter what the origin of the request is. The default setting for proxy.chp and proxy.traefik’s networkPolicy configuration is [http, https], while it is [] for other networkPolicies.

Note that these port names or numbers target a Pod’s port name or number, not a k8s Service’s port name or number.

Default: []

hub.db#

hub.db.type#

Type of database backend to use for the hub database.

The Hub requires a persistent database to function, and this lets you specify where it should be stored.

The various options are:

  1. sqlite-pvc

    Use an sqlite database kept on a persistent volume attached to the hub.

    By default, this disk is created by the cloud provider using dynamic provisioning configured by a storage class. You can customize how this disk is created / attached by setting various properties under hub.db.pvc.

    This is the default setting, and should work well for most cloud provider deployments.

  2. sqlite-memory

    Use an in-memory sqlite database. This should only be used for testing, since the database is erased whenever the hub pod restarts - causing the hub to lose all memory of users who had logged in before.

    When using this for testing, make sure you delete all other objects that the hub has created (such as user pods, user PVCs, etc) every time the hub restarts. Otherwise you might run into errors about duplicate resources.

  3. mysql

    Use an externally hosted mysql database.

    You have to specify an sqlalchemy connection string for the mysql database you want to connect to in hub.db.url if using this option.

    The general format of the connection string is:

    mysql+pymysql://<db-username>:<db-password>@<db-hostname>:<db-port>/<db-name>
    

    The user specified in the connection string must have the rights to create tables in the database specified.

  4. postgres

    Use an externally hosted postgres database.

    You have to specify an sqlalchemy connection string for the postgres database you want to connect to in hub.db.url if using this option.

    The general format of the connection string is:

    postgresql+psycopg2://<db-username>:<db-password>@<db-hostname>:<db-port>/<db-name>
    

    The user specified in the connection string must have the rights to create tables in the database specified.

  5. other

    Use an externally hosted database of some kind other than mysql or postgres.

    When using other, the database password must be passed as part of hub.db.url as hub.db.password will be ignored.

Default: "sqlite-pvc"

hub.db.pvc#

Customize the Persistent Volume Claim used when hub.db.type is sqlite-pvc.

hub.db.pvc.annotations#

Annotations to apply to the PVC containing the sqlite database.

See the Kubernetes documentation for more details about annotations.

hub.db.pvc.selector#

Label selectors to set for the PVC containing the sqlite database.

Useful when you are using a specific PV, and want to bind to that and only that.

See the Kubernetes documentation for more details about using a label selector for what PV to bind to.

hub.db.pvc.storage#

Size of disk to request for the database disk.

Default: "1Gi"

hub.db.pvc.accessModes#

AccessModes contains the desired access modes the volume should have. See the k8s documentation for more information.

Default: ["ReadWriteOnce"]

hub.db.pvc.storageClassName#

Name of the StorageClass required by the claim.

If this is a blank string it will be set to a blank string, while if it is null, it will not be set at all.

hub.db.pvc.subPath#

Path within the volume from which the container’s volume should be mounted. Defaults to “” (volume’s root).

hub.db.upgrade#

Users with external databases need to opt-in for upgrades of the JupyterHub specific database schema if needed as part of a JupyterHub version upgrade.

hub.db.url#

Connection string when hub.db.type is mysql or postgres.

See documentation for hub.db.type for more details on the format of this property.

hub.db.password#

Password for the database when hub.db.type is mysql or postgres.

hub.labels#

Extra labels to add to the hub pod.

See the Kubernetes docs to learn more about labels.

hub.initContainers#

list of initContainers to be run with hub pod. See Kubernetes Docs

hub:
  initContainers:
    - name: init-myservice
      image: busybox:1.28
      command: ['sh', '-c', 'command1']
    - name: init-mydb
      image: busybox:1.28
      command: ['sh', '-c', 'command2']

Default: []

hub.extraEnv#

Extra environment variables that should be set for the hub pod.

Environment variables are usually used to:

  • Pass parameters to some custom code in hub.extraConfig.

  • Configure code running in the hub pod, such as an authenticator or spawner.

String literals with $(ENV_VAR_NAME) will be expanded by Kubelet which is a part of Kubernetes.

hub:
  extraEnv:
    # basic notation (for literal values only)
    MY_ENV_VARS_NAME1: "my env var value 1"

    # explicit notation (the "name" field takes precedence)
    HUB_NAMESPACE:
      name: HUB_NAMESPACE
      valueFrom:
        fieldRef:
          fieldPath: metadata.namespace

    # implicit notation (the "name" field is implied)
    PREFIXED_HUB_NAMESPACE:
      value: "my-prefix-$(HUB_NAMESPACE)"
    SECRET_VALUE:
      valueFrom:
        secretKeyRef:
          name: my-k8s-secret
          key: password

For more information, see the Kubernetes EnvVar specification.

hub.extraConfig#

Arbitrary extra python based configuration that should be in jupyterhub_config.py.

This is the escape hatch - if you want to configure JupyterHub to do something specific that is not present here as an option, you can write the raw Python to do it here.

extraConfig is a dict, so there can be multiple configuration snippets under different names. The configuration sections are run in alphabetical order based on the keys.

Non-exhaustive examples of things you can do here:

  • Subclass authenticator / spawner to do a custom thing

  • Dynamically launch different images for different sets of images

  • Inject an auth token from GitHub authenticator into user pod

  • Anything else you can think of!

Since this is usually a multi-line string, you want to format it using YAML’s | operator.

For example:

hub:
  extraConfig:
    myConfig.py: |
      c.JupyterHub.something = 'something'
      c.Spawner.something_else = 'something else'

Note

No code validation is performed until JupyterHub loads it! If you make a typo here, it will probably manifest itself as the hub pod failing to start up and instead entering an Error state or the subsequent CrashLoopBackoff state.

To make use of your own programs linters etc, it would be useful to not embed Python code inside a YAML file. To do that, consider using hub.extraFiles and mounting a file to /usr/local/etc/jupyterhub/jupyterhub_config.d in order to load your extra configuration logic.

hub.fsGid#

Note

Removed in version 2.0.0. Use hub.podSecurityContext and specify fsGroup instead.

hub.service#

Object to configure the service the JupyterHub will be exposed on by the Kubernetes server.

hub.service.type#

The Kubernetes ServiceType to be used.

The default type is ClusterIP. See the Kubernetes docs to learn more about service types.

Default: "ClusterIP"

hub.service.ports#

Object to configure the ports the hub service will be deployed on.

hub.service.ports.nodePort#

The nodePort to deploy the hub service on.

hub.service.annotations#

Kubernetes annotations to apply to the hub service.

hub.service.extraPorts#

Extra ports to add to the Hub Service object besides hub / 8081. This should be an array that includes name, port, and targetPort. See Multi-port Services for more details.

Default: []

hub.service.loadBalancerIP#

A public IP address the hub Kubernetes service should be exposed on. To expose the hub directly is not recommended. Instead route traffic through the proxy-public service towards the hub.

hub.pdb#

Configure a PodDisruptionBudget for this Deployment.

These are disabled by default for our deployments that don’t support being run in parallel with multiple replicas. Only the user-scheduler currently supports being run in parallel with multiple replicas. If they are enabled for a Deployment with only one replica, they will block kubectl drain of a node for example.

Note that if you aim to block scaling down a node with the hub/proxy/autohttps pod that would cause disruptions of the deployment, then you should instead annotate the pods of the Deployment as described here.

"cluster-autoscaler.kubernetes.io/safe-to-evict": "false"

See the Kubernetes documentation for more details about disruptions.

hub.pdb.enabled#

Decides if a PodDisruptionBudget is created targeting the Deployment’s pods.

Default: false

hub.pdb.maxUnavailable#

The maximum number of pods that can be unavailable during voluntary disruptions.

hub.pdb.minAvailable#

The minimum number of pods required to be available during voluntary disruptions.

Default: 1

hub.existingSecret#

This option allow you to provide the name of an existing k8s Secret to use alongside of the chart managed k8s Secret. The content of this k8s Secret will be merged with the chart managed k8s Secret, giving priority to the self-managed k8s Secret.

Warning

  1. The self managed k8s Secret must mirror the structure in the chart managed secret.

  2. proxy.secretToken (aka. hub.config.ConfigurableHTTPProxy.auth_token) is only read from the chart managed k8s Secret.

hub.nodeSelector#

An object with key value pairs representing labels. K8s Nodes are required to have match all these labels for this Pod to scheduled on them.

disktype: ssd
nodetype: awesome

See the Kubernetes documentation for more details.

hub.tolerations#

Tolerations allow a pod to be scheduled on nodes with taints. These tolerations are additional tolerations to the tolerations common to all pods of a their respective kind (scheduling.corePods.tolerations, scheduling.userPods.tolerations).

Pass this field an array of Toleration objects.

See the Kubernetes docs for more info.

Default: []

hub.activeServerLimit#

JupyterHub native configuration, see the JupyterHub documentation for more information.

hub.allowNamedServers#

JupyterHub native configuration, see the JupyterHub documentation for more information.

Default: false

hub.annotations#

K8s annotations for the hub pod.

hub.authenticatePrometheus#

JupyterHub native configuration, see the JupyterHub documentation for more information.

hub.concurrentSpawnLimit#

JupyterHub native configuration, see the JupyterHub documentation for more information.

Default: 64

hub.consecutiveFailureLimit#

JupyterHub native configuration, see the JupyterHub documentation for more information.

Default: 5

hub.podSecurityContext#

A k8s native specification of the pod’s security context, see the documentation for details.

hub.containerSecurityContext#

A k8s native specification of the container’s security context, see the documentation for details.

hub.deploymentStrategy#

hub.deploymentStrategy.rollingUpdate#

hub.deploymentStrategy.type#

JupyterHub does not support running in parallel, due to this we default to using a deployment strategy of Recreate.

Default: "Recreate"

hub.extraContainers#

Additional containers for the Pod. Use a k8s native syntax.

Default: []

hub.extraVolumeMounts#

Additional volume mounts for the Container. Use a k8s native syntax.

Default: []

hub.extraVolumes#

Additional volumes for the Pod. Use a k8s native syntax.

Default: []

hub.livenessProbe#

hub.readinessProbe#

hub.namedServerLimitPerUser#

JupyterHub native configuration, see the JupyterHub documentation for more information.

hub.redirectToServer#

JupyterHub native configuration, see the JupyterHub documentation for more information.

hub.resources#

A k8s native specification of resources, see the documentation.

hub.lifecycle#

A k8s native specification of lifecycle hooks on the container, see the documentation.

hub.lifecycle.postStart#

hub.lifecycle.preStop#

hub.services#

This is where you register JupyterHub services. For details on how to configure these services in this Helm chart just keep reading but for details on services themselves instead read JupyterHub’s documentation.

Note

Only a selection of JupyterHub’s configuration options that can be configured for a service are documented below. All configuration set here will be applied even if this Helm chart doesn’t recognize it.

JupyterHub’s native configuration accepts a list of service objects, this Helm chart only accept a dictionary where each key represents the name of a service and the value is the actual service objects.

When configuring JupyterHub services via this Helm chart, the name field can be omitted as it can be implied by the dictionary key. Further, the api_token field can be omitted as it will be automatically generated as of version 1.1.0 of this Helm chart.

If you have an external service that needs to access the automatically generated api_token for the service, you can access it from the hub k8s Secret part of this Helm chart under the key hub.services.my-service-config-key.apiToken.

Here is an example configuration of two services where the first explicitly sets a name and api_token, while the second omits those and lets the name be implied from the key name and the api_token be automatically generated.

hub:
  services:
    my-service-1:
      admin: true
      name: my-explicitly-set-service-name
      api_token: my-explicitly-set-api_token

    # the name of the following service will be my-service-2
    # the api_token of the following service will be generated
    my-service-2: {}

If you develop a Helm chart depending on the JupyterHub Helm chart and want to let some Pod’s environment variable be populated with the api_token of a service registered like above, then do something along these lines.

# ... container specification of a pod ...
env:
  - name: MY_SERVICE_1_API_TOKEN
    valueFrom:
      secretKeyRef:
        # Don't hardcode the name, use the globally accessible
        # named templates part of the JupyterHub Helm chart.
        name: {{ include "jupyterhub.hub.fullname" . }}
        # Note below the use of the configuration key my-service-1
        # rather than the explicitly set service name.
        key: hub.services.my-service-1.apiToken

hub.services.name#

The name can be implied via the key name under which this service is configured, and is due to that allowed to be omitted in this Helm chart configuration of JupyterHub.

hub.services.admin#

hub.services.command#

hub.services.url#

hub.services.api_token#

The api_token will be automatically generated if not explicitly set. It will also be exposed in via a k8s Secret part of this Helm chart under a specific key.

See the documentation under hub.services for details about this.

hub.services.apiToken#

An alias for api_token provided for backward compatibility by the JupyterHub Helm chart that will be transformed to api_token.

hub.loadRoles#

This is where you should define JupyterHub roles and apply them to JupyterHub users, groups, and services to grant them additional permissions as defined in JupyterHub’s RBAC system.

Complement this documentation with JupyterHub’s documentation about load_roles.

Note that while JupyterHub’s native configuration load_roles accepts a list of role objects, this Helm chart only accepts a dictionary where each key represents the name of a role and the value is the actual role object.

hub:
  loadRoles:
    teacher:
      description: Access to users' information and group membership

      # this role provides permissions to...
      scopes: [users, groups]

      # this role will be assigned to...
      users: [erik]
      services: [grading-service]
      groups: [teachers]

When configuring JupyterHub roles via this Helm chart, the name field can be omitted as it can be implied by the dictionary key.

hub.shutdownOnLogout#

JupyterHub native configuration, see the JupyterHub documentation for more information.

hub.templatePaths#

JupyterHub native configuration, see the JupyterHub documentation for more information.

Default: []

hub.templateVars#

JupyterHub native configuration, see the JupyterHub documentation for more information.

hub.serviceAccount#

Configuration for a k8s ServiceAccount dedicated for use by the specific pod which this configuration is nested under.

hub.serviceAccount.create#

Whether or not to create the ServiceAccount resource.

Default: true

hub.serviceAccount.name#

This configuration serves multiple purposes:

  • It will be the serviceAccountName referenced by related Pods.

  • If create is set, the created ServiceAccount resource will be named like this.

  • If rbac.create is set, the associated (Cluster)RoleBindings will bind to this name.

If not explicitly provided, a default name will be used.

hub.serviceAccount.annotations#

Kubernetes annotations to apply to the k8s ServiceAccount.

hub.extraPodSpec#

Arbitrary extra k8s pod specification as a YAML object. The default value of this setting is an empty object, i.e. no extra configuration. The value of this property is augmented to the pod specification as-is.

This is a powerful tool for expert k8s administrators with advanced configuration requirements. This setting should only be used for configuration that cannot be accomplished through the other settings. Misusing this setting can break your deployment and/or compromise your system security.

This is one of four related settings for inserting arbitrary pod specification:

  1. hub.extraPodSpec

  2. proxy.chp.extraPodSpec

  3. proxy.traefik.extraPodSpec

  4. scheduling.userScheduler.extraPodSpec

One real-world use of these settings is to enable host networking. For example, to configure host networking for the hub pod, add the following to your helm configuration values:

hub:
  extraPodSpec:
    hostNetwork: true
    dnsPolicy: ClusterFirstWithHostNet

Likewise, to configure host networking for the proxy pod, add the following:

proxy:
  chp:
    extraPodSpec:
      hostNetwork: true
      dnsPolicy: ClusterFirstWithHostNet

N.B. Host networking has special security implications and can easily break your deployment. This is an example—not an endorsement.

See PodSpec for the latest pod resource specification.

proxy#

proxy.chp#

Configure the configurable-http-proxy (chp) pod managed by jupyterhub to route traffic both to itself and to user pods.

proxy.chp.revisionHistoryLimit#

Configures the resource’s spec.revisionHistoryLimit. This is available for Deployment, StatefulSet, and DaemonSet resources.

See the Kubernetes docs for more info.

proxy.chp.networkPolicy#

This configuration regards the creation and configuration of a k8s NetworkPolicy resource.

proxy.chp.networkPolicy.enabled#

Toggle the creation of the NetworkPolicy resource targeting this pod, and by doing so, restricting its communication to only what is explicitly allowed in the NetworkPolicy.

Default: true

proxy.chp.networkPolicy.ingress#

Additional ingress rules to add besides those that are required for core functionality.

Default: []

proxy.chp.networkPolicy.egress#

Additional egress rules to add besides those that are required for core functionality and those added via .egressAllowRules.

Changed in version 2.0.0: The default value changed from providing one very permissive rule allowing all egress to providing no rule. The permissive rule is still provided via .egressAllowRules set to true though.

As an example, below is a configuration that disables the more broadly permissive .privateIPs egress allow rule for the hub pod, and instead provides tightly scoped permissions to access a specific k8s local service as identified by pod labels.

hub:
  networkPolicy:
    egressAllowRules:
      privateIPs: false
    egress:
      - to:
          - podSelector:
              matchLabels:
                app: my-k8s-local-service
        ports:
          - protocol: TCP
            port: 5978

Default: []

proxy.chp.networkPolicy.egressAllowRules#

This is a set of predefined rules that when enabled will be added to the NetworkPolicy list of egress rules.

The resulting egress rules will be a composition of:

  • rules specific for the respective pod(s) function within the Helm chart

  • rules based on enabled egressAllowRules flags

  • rules explicitly specified by the user

Note

Each flag under this configuration will not render into a dedicated rule in the NetworkPolicy resource, but instead combine with the other flags to a reduced set of rules to avoid a performance penalty.

New in version 2.0.0.

proxy.chp.networkPolicy.egressAllowRules.cloudMetadataServer#

Defaults to false for singleuser servers, but to true for all other network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the cloud metadata server.

Note that the nonPrivateIPs rule is allowing all non Private IP ranges but makes an exception for the cloud metadata server, leaving this as the definitive configuration to allow access to the cloud metadata server.

Changed in version 3.0.0: This configuration is not allowed to be configured true at the same time as singleuser.cloudMetadata.blockWithIptables to avoid an ambiguous configuration.

Default: true

proxy.chp.networkPolicy.egressAllowRules.dnsPortsCloudMetadataServer#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the cloud metadata server via port 53.

Relying on this rule for the singleuser config should go hand in hand with disabling singleuser.cloudMetadata.blockWithIptables to avoid an ambiguous configuration.

Known situations when this rule can be relevant:

  • In GKE clusters with Cloud DNS that is reached at the cloud metadata server’s non-private IP.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

New in version 3.0.0.

Default: true

proxy.chp.networkPolicy.egressAllowRules.dnsPortsKubeSystemNamespace#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to pods in the kube-system namespace via port 53.

Known situations when this rule can be relevant:

  • GKE, EKS, AKS, and other clusters relying directly on kube-dns or coredns pods in the kube-system namespace.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

New in version 3.0.0.

Default: true

proxy.chp.networkPolicy.egressAllowRules.dnsPortsPrivateIPs#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to private IPs via port 53.

Known situations when this rule can be relevant:

  • GKE clusters relying on a DNS server indirectly via a a node local DNS cache at an unknown private IP.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

Warning

This rule is not expected to work in clusters relying on Cilium to enforce the NetworkPolicy rules (includes GKE clusters with Dataplane v2), this is due to a known limitation.

Default: true

proxy.chp.networkPolicy.egressAllowRules.nonPrivateIPs#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the non-private IP ranges with the exception of the cloud metadata server. This means respective pod(s) can establish connections to the internet but not (say) an unsecured prometheus server running in the same cluster.

Default: true

proxy.chp.networkPolicy.egressAllowRules.privateIPs#

Defaults to false for singleuser servers, but to true for all other network policies.

Private IPs refer to the IP ranges 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16.

When enabled this rule allows the respective pod(s) to establish outbound connections to the internal k8s cluster. This means users can access the internet but not (say) an unsecured prometheus server running in the same cluster.

Since not all workloads in the k8s cluster may have NetworkPolicies setup to restrict their incoming connections, having this set to false can be a good defense against malicious intent from someone in control of software in these pods.

If possible, try to avoid setting this to true as it gives broad permissions that could be specified more directly via the .egress.

Warning

This rule is not expected to work in clusters relying on Cilium to enforce the NetworkPolicy rules (includes GKE clusters with Dataplane v2), this is due to a known limitation.

Default: true

proxy.chp.networkPolicy.interNamespaceAccessLabels#

This configuration option determines if both namespaces and pods in other namespaces, that have specific access labels, should be accepted to allow ingress (set to accept), or, if the labels are to be ignored when applied outside the local namespace (set to ignore).

The available access labels for respective NetworkPolicy resources are:

  • hub.jupyter.org/network-access-hub: "true" (hub)

  • hub.jupyter.org/network-access-proxy-http: "true" (proxy.chp, proxy.traefik)

  • hub.jupyter.org/network-access-proxy-api: "true" (proxy.chp)

  • hub.jupyter.org/network-access-singleuser: "true" (singleuser)

Default: "ignore"

proxy.chp.networkPolicy.allowedIngressPorts#

A rule to allow ingress on these ports will be added no matter what the origin of the request is. The default setting for proxy.chp and proxy.traefik’s networkPolicy configuration is [http, https], while it is [] for other networkPolicies.

Note that these port names or numbers target a Pod’s port name or number, not a k8s Service’s port name or number.

Default: ["http", "https"]

proxy.chp.extraCommandLineFlags#

A list of strings to be added as command line options when starting configurable-http-proxy that will be expanded with Helm’s template function tpl which can render Helm template logic inside curly braces ({{ ... }}).

proxy:
  chp:
    extraCommandLineFlags:
      - "--auto-rewrite"
      - "--custom-header={{ .Values.custom.myStuff }}"

Note that these will be appended last, and if you provide the same flag twice, the last flag will be used, which mean you can override the default flag values as well.

Default: []

proxy.chp.extraEnv#

Extra environment variables that should be set for the chp pod.

Environment variables are usually used here to:

  • override HUB_SERVICE_PORT or HUB_SERVICE_HOST default values

  • set CONFIGPROXY_SSL_KEY_PASSPHRASE for setting passphrase of SSL keys

String literals with $(ENV_VAR_NAME) will be expanded by Kubelet which is a part of Kubernetes.

proxy:
  chp:
    extraEnv:
      # basic notation (for literal values only)
      MY_ENV_VARS_NAME1: "my env var value 1"

      # explicit notation (the "name" field takes precedence)
      CHP_NAMESPACE:
        name: CHP_NAMESPACE
        valueFrom:
          fieldRef:
            fieldPath: metadata.namespace

      # implicit notation (the "name" field is implied)
      PREFIXED_CHP_NAMESPACE:
        value: "my-prefix-$(CHP_NAMESPACE)"
      SECRET_VALUE:
        valueFrom:
          secretKeyRef:
            name: my-k8s-secret
            key: password

For more information, see the Kubernetes EnvVar specification.

proxy.chp.pdb#

Configure a PodDisruptionBudget for this Deployment.

These are disabled by default for our deployments that don’t support being run in parallel with multiple replicas. Only the user-scheduler currently supports being run in parallel with multiple replicas. If they are enabled for a Deployment with only one replica, they will block kubectl drain of a node for example.

Note that if you aim to block scaling down a node with the hub/proxy/autohttps pod that would cause disruptions of the deployment, then you should instead annotate the pods of the Deployment as described here.

"cluster-autoscaler.kubernetes.io/safe-to-evict": "false"

See the Kubernetes documentation for more details about disruptions.

proxy.chp.pdb.enabled#

Decides if a PodDisruptionBudget is created targeting the Deployment’s pods.

Default: false

proxy.chp.pdb.maxUnavailable#

The maximum number of pods that can be unavailable during voluntary disruptions.

proxy.chp.pdb.minAvailable#

The minimum number of pods required to be available during voluntary disruptions.

Default: 1

proxy.chp.nodeSelector#

An object with key value pairs representing labels. K8s Nodes are required to have match all these labels for this Pod to scheduled on them.

disktype: ssd
nodetype: awesome

See the Kubernetes documentation for more details.

proxy.chp.tolerations#

Tolerations allow a pod to be scheduled on nodes with taints. These tolerations are additional tolerations to the tolerations common to all pods of a their respective kind (scheduling.corePods.tolerations, scheduling.userPods.tolerations).

Pass this field an array of Toleration objects.

See the Kubernetes docs for more info.

Default: []

proxy.chp.containerSecurityContext#

A k8s native specification of the container’s security context, see the documentation for details.

proxy.chp.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

proxy.chp.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "quay.io/jupyterhub/configurable-http-proxy"

proxy.chp.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a

Default: "4.6.1"

proxy.chp.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

proxy.chp.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

proxy.chp.livenessProbe#

proxy.chp.readinessProbe#

proxy.chp.resources#

A k8s native specification of resources, see the documentation.

proxy.chp.defaultTarget#

Override the URL for the default routing target for the proxy. Defaults to JupyterHub itself. This will generally only have an effect while JupyterHub is not running, as JupyterHub adds itself as the default target after it starts.

proxy.chp.errorTarget#

Override the URL for the error target for the proxy. Defaults to JupyterHub itself. Useful to reduce load on the Hub or produce more informative error messages than the Hub’s default, e.g. in highly customized deployments such as BinderHub. See Configurable HTTP Proxy for details on implementing an error target.

proxy.chp.extraPodSpec#

Arbitrary extra k8s pod specification as a YAML object. The default value of this setting is an empty object, i.e. no extra configuration. The value of this property is augmented to the pod specification as-is.

This is a powerful tool for expert k8s administrators with advanced configuration requirements. This setting should only be used for configuration that cannot be accomplished through the other settings. Misusing this setting can break your deployment and/or compromise your system security.

This is one of four related settings for inserting arbitrary pod specification:

  1. hub.extraPodSpec

  2. proxy.chp.extraPodSpec

  3. proxy.traefik.extraPodSpec

  4. scheduling.userScheduler.extraPodSpec

One real-world use of these settings is to enable host networking. For example, to configure host networking for the hub pod, add the following to your helm configuration values:

hub:
  extraPodSpec:
    hostNetwork: true
    dnsPolicy: ClusterFirstWithHostNet

Likewise, to configure host networking for the proxy pod, add the following:

proxy:
  chp:
    extraPodSpec:
      hostNetwork: true
      dnsPolicy: ClusterFirstWithHostNet

N.B. Host networking has special security implications and can easily break your deployment. This is an example—not an endorsement.

See PodSpec for the latest pod resource specification.

proxy.secretToken#

Note

As of version 1.0.0 this will automatically be generated and there is no need to set it manually.

If you wish to reset a generated key, you can use kubectl edit on the k8s Secret typically named hub and remove the hub.config.ConfigurableHTTPProxy.auth_token entry in the k8s Secret, then perform a new helm upgrade.

A 32-byte cryptographically secure randomly generated string used to secure communications between the hub pod and the proxy pod running a configurable-http-proxy instance.

# to generate a value, run
openssl rand -hex 32

Changing this value will cause the proxy and hub pods to restart. It is good security practice to rotate these values over time. If this secret leaks, immediately change it to something else, or user data can be compromised.

proxy.service#

Configuration of the k8s Service proxy-public which either will point to the autohttps pod running Traefik for TLS termination, or the proxy pod running ConfigurableHTTPProxy. Incoming traffic from users on the internet should always go through this k8s Service.

When this service targets the autohttps pod which then routes to the proxy pod, a k8s Service named proxy-http will be added targeting the proxy pod and only accepting HTTP traffic on port 80.

proxy.service.type#

Default LoadBalancer. See the Kubernetes docs to learn more about service types.

Default: "LoadBalancer"

proxy.service.labels#

Extra labels to add to the proxy service.

See the Kubernetes docs to learn more about labels.

proxy.service.annotations#

Annotations to apply to the service that is exposing the proxy.

See the Kubernetes documentation for more details about annotations.

proxy.service.nodePorts#

Object to set NodePorts to expose the service on for http and https.

See the Kubernetes documentation for more details about NodePorts.

proxy.service.nodePorts.http#

The HTTP port the proxy-public service should be exposed on.

proxy.service.nodePorts.https#

The HTTPS port the proxy-public service should be exposed on.

proxy.service.disableHttpPort#

Default false.

If true, port 80 for incoming HTTP traffic will no longer be exposed. This should not be used with proxy.https.type=letsencrypt or proxy.https.enabled=false as it would remove the only exposed port.

Default: false

proxy.service.extraPorts#

Extra ports the k8s Service should accept incoming traffic on, which will be redirected to either the autohttps pod (treafik) or the proxy pod (chp).

See the Kubernetes documentation for the structure of the items in this list.

Default: []

proxy.service.loadBalancerIP#

The public IP address the proxy-public Kubernetes service should be exposed on. This entry will end up at the configurable proxy server that JupyterHub manages, which will direct traffic to user pods at the /user path and the hub pod at the /hub path.

Set this if you want to use a fixed external IP address instead of a dynamically acquired one. This is relevant if you have a domain name that you want to point to a specific IP and want to ensure it doesn’t change.

proxy.service.loadBalancerSourceRanges#

A list of IP CIDR ranges that are allowed to access the load balancer service. Defaults to allowing everyone to access it.

Default: []

proxy.https#

Object for customizing the settings for HTTPS used by the JupyterHub’s proxy. For more information on configuring HTTPS for your JupyterHub, see the HTTPS section in our security guide

proxy.https.enabled#

Indicator to set whether HTTPS should be enabled or not on the proxy. Defaults to true if the https object is provided.

Default: false

proxy.https.type#

The type of HTTPS encryption that is used. Decides on which ports and network policies are used for communication via HTTPS. Setting this to secret sets the type to manual HTTPS with a secret that has to be provided in the https.secret object. Defaults to letsencrypt.

Default: "letsencrypt"

proxy.https.letsencrypt#

proxy.https.letsencrypt.contactEmail#

The contact email to be used for automatically provisioned HTTPS certificates by Let’s Encrypt. For more information see Set up automatic HTTPS. Required for automatic HTTPS.

proxy.https.letsencrypt.acmeServer#

Let’s Encrypt is one of various ACME servers that can provide a certificate, and by default their production server is used.

Let’s Encrypt staging: https://acme-staging-v02.api.letsencrypt.org/directory Let’s Encrypt production: acmeServer: https://acme-v02.api.letsencrypt.org/directory

Default: "https://acme-v02.api.letsencrypt.org/directory"

proxy.https.manual#

Object for providing own certificates for manual HTTPS configuration. To be provided when setting https.type to manual. See Set up manual HTTPS

proxy.https.manual.key#

The RSA private key to be used for HTTPS. To be provided in the form of

key: |
  -----BEGIN RSA PRIVATE KEY-----
  ...
  -----END RSA PRIVATE KEY-----
proxy.https.manual.cert#

The certificate to be used for HTTPS. To be provided in the form of

cert: |
  -----BEGIN CERTIFICATE-----
  ...
  -----END CERTIFICATE-----

proxy.https.secret#

Secret to be provided when setting https.type to secret.

proxy.https.secret.name#

Name of the secret

proxy.https.secret.key#

Path to the private key to be used for HTTPS. Example: 'tls.key'

Default: "tls.key"

proxy.https.secret.crt#

Path to the certificate to be used for HTTPS. Example: 'tls.crt'

Default: "tls.crt"

proxy.https.hosts#

You domain in list form. Required for automatic HTTPS. See Set up automatic HTTPS. To be provided like:

hosts:
  - <your-domain-name>

Default: []

proxy.traefik#

Configure the traefik proxy used to terminate TLS when ‘autohttps’ is enabled

proxy.traefik.revisionHistoryLimit#

Configures the resource’s spec.revisionHistoryLimit. This is available for Deployment, StatefulSet, and DaemonSet resources.

See the Kubernetes docs for more info.

proxy.traefik.labels#

Extra labels to add to the traefik pod.

See the Kubernetes docs to learn more about labels.

proxy.traefik.networkPolicy#

This configuration regards the creation and configuration of a k8s NetworkPolicy resource.

proxy.traefik.networkPolicy.enabled#

Toggle the creation of the NetworkPolicy resource targeting this pod, and by doing so, restricting its communication to only what is explicitly allowed in the NetworkPolicy.

Default: true

proxy.traefik.networkPolicy.ingress#

Additional ingress rules to add besides those that are required for core functionality.

Default: []

proxy.traefik.networkPolicy.egress#

Additional egress rules to add besides those that are required for core functionality and those added via .egressAllowRules.

Changed in version 2.0.0: The default value changed from providing one very permissive rule allowing all egress to providing no rule. The permissive rule is still provided via .egressAllowRules set to true though.

As an example, below is a configuration that disables the more broadly permissive .privateIPs egress allow rule for the hub pod, and instead provides tightly scoped permissions to access a specific k8s local service as identified by pod labels.

hub:
  networkPolicy:
    egressAllowRules:
      privateIPs: false
    egress:
      - to:
          - podSelector:
              matchLabels:
                app: my-k8s-local-service
        ports:
          - protocol: TCP
            port: 5978

Default: []

proxy.traefik.networkPolicy.egressAllowRules#

This is a set of predefined rules that when enabled will be added to the NetworkPolicy list of egress rules.

The resulting egress rules will be a composition of:

  • rules specific for the respective pod(s) function within the Helm chart

  • rules based on enabled egressAllowRules flags

  • rules explicitly specified by the user

Note

Each flag under this configuration will not render into a dedicated rule in the NetworkPolicy resource, but instead combine with the other flags to a reduced set of rules to avoid a performance penalty.

New in version 2.0.0.

proxy.traefik.networkPolicy.egressAllowRules.cloudMetadataServer#

Defaults to false for singleuser servers, but to true for all other network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the cloud metadata server.

Note that the nonPrivateIPs rule is allowing all non Private IP ranges but makes an exception for the cloud metadata server, leaving this as the definitive configuration to allow access to the cloud metadata server.

Changed in version 3.0.0: This configuration is not allowed to be configured true at the same time as singleuser.cloudMetadata.blockWithIptables to avoid an ambiguous configuration.

Default: true

proxy.traefik.networkPolicy.egressAllowRules.dnsPortsCloudMetadataServer#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the cloud metadata server via port 53.

Relying on this rule for the singleuser config should go hand in hand with disabling singleuser.cloudMetadata.blockWithIptables to avoid an ambiguous configuration.

Known situations when this rule can be relevant:

  • In GKE clusters with Cloud DNS that is reached at the cloud metadata server’s non-private IP.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

New in version 3.0.0.

Default: true

proxy.traefik.networkPolicy.egressAllowRules.dnsPortsKubeSystemNamespace#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to pods in the kube-system namespace via port 53.

Known situations when this rule can be relevant:

  • GKE, EKS, AKS, and other clusters relying directly on kube-dns or coredns pods in the kube-system namespace.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

New in version 3.0.0.

Default: true

proxy.traefik.networkPolicy.egressAllowRules.dnsPortsPrivateIPs#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to private IPs via port 53.

Known situations when this rule can be relevant:

  • GKE clusters relying on a DNS server indirectly via a a node local DNS cache at an unknown private IP.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

Warning

This rule is not expected to work in clusters relying on Cilium to enforce the NetworkPolicy rules (includes GKE clusters with Dataplane v2), this is due to a known limitation.

Default: true

proxy.traefik.networkPolicy.egressAllowRules.nonPrivateIPs#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the non-private IP ranges with the exception of the cloud metadata server. This means respective pod(s) can establish connections to the internet but not (say) an unsecured prometheus server running in the same cluster.

Default: true

proxy.traefik.networkPolicy.egressAllowRules.privateIPs#

Defaults to false for singleuser servers, but to true for all other network policies.

Private IPs refer to the IP ranges 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16.

When enabled this rule allows the respective pod(s) to establish outbound connections to the internal k8s cluster. This means users can access the internet but not (say) an unsecured prometheus server running in the same cluster.

Since not all workloads in the k8s cluster may have NetworkPolicies setup to restrict their incoming connections, having this set to false can be a good defense against malicious intent from someone in control of software in these pods.

If possible, try to avoid setting this to true as it gives broad permissions that could be specified more directly via the .egress.

Warning

This rule is not expected to work in clusters relying on Cilium to enforce the NetworkPolicy rules (includes GKE clusters with Dataplane v2), this is due to a known limitation.

Default: true

proxy.traefik.networkPolicy.interNamespaceAccessLabels#

This configuration option determines if both namespaces and pods in other namespaces, that have specific access labels, should be accepted to allow ingress (set to accept), or, if the labels are to be ignored when applied outside the local namespace (set to ignore).

The available access labels for respective NetworkPolicy resources are:

  • hub.jupyter.org/network-access-hub: "true" (hub)

  • hub.jupyter.org/network-access-proxy-http: "true" (proxy.chp, proxy.traefik)

  • hub.jupyter.org/network-access-proxy-api: "true" (proxy.chp)

  • hub.jupyter.org/network-access-singleuser: "true" (singleuser)

Default: "ignore"

proxy.traefik.networkPolicy.allowedIngressPorts#

A rule to allow ingress on these ports will be added no matter what the origin of the request is. The default setting for proxy.chp and proxy.traefik’s networkPolicy configuration is [http, https], while it is [] for other networkPolicies.

Note that these port names or numbers target a Pod’s port name or number, not a k8s Service’s port name or number.

Default: ["http", "https"]

proxy.traefik.extraInitContainers#

list of extraInitContainers to be run with traefik pod, after the containers set in the chart. See Kubernetes Docs

proxy:
  traefik:
    extraInitContainers:
      - name: init-myservice
        image: busybox:1.28
        command: ['sh', '-c', 'command1']
      - name: init-mydb
        image: busybox:1.28
        command: ['sh', '-c', 'command2']

Default: []

proxy.traefik.extraEnv#

Extra environment variables that should be set for the traefik pod.

Environment Variables here may be used to configure traefik.

String literals with $(ENV_VAR_NAME) will be expanded by Kubelet which is a part of Kubernetes.

proxy:
  traefik:
    extraEnv:
      # basic notation (for literal values only)
      MY_ENV_VARS_NAME1: "my env var value 1"

      # explicit notation (the "name" field takes precedence)
      TRAEFIK_NAMESPACE:
        name: TRAEFIK_NAMESPACE
        valueFrom:
          fieldRef:
            fieldPath: metadata.namespace

      # implicit notation (the "name" field is implied)
      PREFIXED_TRAEFIK_NAMESPACE:
        value: "my-prefix-$(TRAEFIK_NAMESPACE)"
      SECRET_VALUE:
        valueFrom:
          secretKeyRef:
            name: my-k8s-secret
            key: password

For more information, see the Kubernetes EnvVar specification.

proxy.traefik.pdb#

Configure a PodDisruptionBudget for this Deployment.

These are disabled by default for our deployments that don’t support being run in parallel with multiple replicas. Only the user-scheduler currently supports being run in parallel with multiple replicas. If they are enabled for a Deployment with only one replica, they will block kubectl drain of a node for example.

Note that if you aim to block scaling down a node with the hub/proxy/autohttps pod that would cause disruptions of the deployment, then you should instead annotate the pods of the Deployment as described here.

"cluster-autoscaler.kubernetes.io/safe-to-evict": "false"

See the Kubernetes documentation for more details about disruptions.

proxy.traefik.pdb.enabled#

Decides if a PodDisruptionBudget is created targeting the Deployment’s pods.

Default: false

proxy.traefik.pdb.maxUnavailable#

The maximum number of pods that can be unavailable during voluntary disruptions.

proxy.traefik.pdb.minAvailable#

The minimum number of pods required to be available during voluntary disruptions.

Default: 1

proxy.traefik.nodeSelector#

An object with key value pairs representing labels. K8s Nodes are required to have match all these labels for this Pod to scheduled on them.

disktype: ssd
nodetype: awesome

See the Kubernetes documentation for more details.

proxy.traefik.tolerations#

Tolerations allow a pod to be scheduled on nodes with taints. These tolerations are additional tolerations to the tolerations common to all pods of a their respective kind (scheduling.corePods.tolerations, scheduling.userPods.tolerations).

Pass this field an array of Toleration objects.

See the Kubernetes docs for more info.

Default: []

proxy.traefik.containerSecurityContext#

A k8s native specification of the container’s security context, see the documentation for details.

proxy.traefik.extraDynamicConfig#

This refers to traefik’s post-startup configuration.

This Helm chart already provide such configuration, so this is a place where you can merge in additional configuration. If you are about to use this configuration, you may want to inspect the default configuration declared here.

proxy.traefik.extraPorts#

Extra ports for the traefik container within the autohttps pod that you would like to expose, formatted in a k8s native way.

Default: []

proxy.traefik.extraStaticConfig#

This refers to traefik’s startup configuration.

This Helm chart already provide such configuration, so this is a place where you can merge in additional configuration. If you are about to use this configuration, you may want to inspect the default configuration declared here.

proxy.traefik.extraVolumes#

Additional volumes for the Pod. Use a k8s native syntax.

Default: []

proxy.traefik.extraVolumeMounts#

Additional volume mounts for the Container. Use a k8s native syntax.

Default: []

proxy.traefik.hsts#

This section regards a HTTP Strict-Transport-Security (HSTS) response header. It can act as a request for a visiting web browsers to enforce HTTPS on their end in for a given time into the future, and optionally also for future requests to subdomains.

These settings relate to traefik configuration which we use as a TLS termination proxy.

See Mozilla’s documentation for more information.

proxy.traefik.hsts.includeSubdomains#

Default: false

proxy.traefik.hsts.maxAge#

Default: 15724800

proxy.traefik.hsts.preload#

Default: false

proxy.traefik.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

proxy.traefik.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "traefik"

proxy.traefik.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a

Default: "v2.11.0"

proxy.traefik.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

proxy.traefik.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

proxy.traefik.resources#

A k8s native specification of resources, see the documentation.

proxy.traefik.serviceAccount#

Configuration for a k8s ServiceAccount dedicated for use by the specific pod which this configuration is nested under.

proxy.traefik.serviceAccount.create#

Whether or not to create the ServiceAccount resource.

Default: true

proxy.traefik.serviceAccount.name#

This configuration serves multiple purposes:

  • It will be the serviceAccountName referenced by related Pods.

  • If create is set, the created ServiceAccount resource will be named like this.

  • If rbac.create is set, the associated (Cluster)RoleBindings will bind to this name.

If not explicitly provided, a default name will be used.

proxy.traefik.serviceAccount.annotations#

Kubernetes annotations to apply to the k8s ServiceAccount.

proxy.traefik.extraPodSpec#

Arbitrary extra k8s pod specification as a YAML object. The default value of this setting is an empty object, i.e. no extra configuration. The value of this property is augmented to the pod specification as-is.

This is a powerful tool for expert k8s administrators with advanced configuration requirements. This setting should only be used for configuration that cannot be accomplished through the other settings. Misusing this setting can break your deployment and/or compromise your system security.

This is one of four related settings for inserting arbitrary pod specification:

  1. hub.extraPodSpec

  2. proxy.chp.extraPodSpec

  3. proxy.traefik.extraPodSpec

  4. scheduling.userScheduler.extraPodSpec

One real-world use of these settings is to enable host networking. For example, to configure host networking for the hub pod, add the following to your helm configuration values:

hub:
  extraPodSpec:
    hostNetwork: true
    dnsPolicy: ClusterFirstWithHostNet

Likewise, to configure host networking for the proxy pod, add the following:

proxy:
  chp:
    extraPodSpec:
      hostNetwork: true
      dnsPolicy: ClusterFirstWithHostNet

N.B. Host networking has special security implications and can easily break your deployment. This is an example—not an endorsement.

See PodSpec for the latest pod resource specification.

proxy.labels#

K8s labels for the proxy pod.

Note

For consistency, this should really be located under proxy.chp.labels but isn’t for historical reasons.

proxy.annotations#

K8s annotations for the proxy pod.

Note

For consistency, this should really be located under proxy.chp.annotations but isn’t for historical reasons.

proxy.deploymentStrategy#

proxy.deploymentStrategy.rollingUpdate#

proxy.deploymentStrategy.type#

While the proxy pod running configurable-http-proxy could run in parallel, two instances running in parallel wouldn’t both receive updates from JupyterHub regarding how it should route traffic. Due to this we default to using a deployment strategy of Recreate instead of RollingUpdate.

Default: "Recreate"

proxy.secretSync#

This configuration section refers to configuration of the sidecar container in the autohttps pod running next to its traefik container responsible for TLS termination.

The purpose of this container is to store away and load TLS certificates from a k8s Secret. The TLS certificates are acquired by the ACME client (LEGO) that is running within the traefik container, where traefik is using them for TLS termination.

proxy.secretSync.containerSecurityContext#

A k8s native specification of the container’s security context, see the documentation for details.

proxy.secretSync.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

proxy.secretSync.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "quay.io/jupyterhub/k8s-secret-sync"

proxy.secretSync.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a
proxy.secretSync.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

proxy.secretSync.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

proxy.secretSync.resources#

A k8s native specification of resources, see the documentation.

singleuser#

Options for customizing the environment that is provided to the users after they log in.

singleuser.networkPolicy#

This configuration regards the creation and configuration of a k8s NetworkPolicy resource.

singleuser.networkPolicy.enabled#

Toggle the creation of the NetworkPolicy resource targeting this pod, and by doing so, restricting its communication to only what is explicitly allowed in the NetworkPolicy.

Default: true

singleuser.networkPolicy.ingress#

Additional ingress rules to add besides those that are required for core functionality.

Default: []

singleuser.networkPolicy.egress#

Additional egress rules to add besides those that are required for core functionality and those added via .egressAllowRules.

Changed in version 2.0.0: The default value changed from providing one very permissive rule allowing all egress to providing no rule. The permissive rule is still provided via .egressAllowRules set to true though.

As an example, below is a configuration that disables the more broadly permissive .privateIPs egress allow rule for the hub pod, and instead provides tightly scoped permissions to access a specific k8s local service as identified by pod labels.

hub:
  networkPolicy:
    egressAllowRules:
      privateIPs: false
    egress:
      - to:
          - podSelector:
              matchLabels:
                app: my-k8s-local-service
        ports:
          - protocol: TCP
            port: 5978

Default: []

singleuser.networkPolicy.egressAllowRules#

This is a set of predefined rules that when enabled will be added to the NetworkPolicy list of egress rules.

The resulting egress rules will be a composition of:

  • rules specific for the respective pod(s) function within the Helm chart

  • rules based on enabled egressAllowRules flags

  • rules explicitly specified by the user

Note

Each flag under this configuration will not render into a dedicated rule in the NetworkPolicy resource, but instead combine with the other flags to a reduced set of rules to avoid a performance penalty.

New in version 2.0.0.

singleuser.networkPolicy.egressAllowRules.cloudMetadataServer#

Defaults to false for singleuser servers, but to true for all other network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the cloud metadata server.

Note that the nonPrivateIPs rule is allowing all non Private IP ranges but makes an exception for the cloud metadata server, leaving this as the definitive configuration to allow access to the cloud metadata server.

Changed in version 3.0.0: This configuration is not allowed to be configured true at the same time as singleuser.cloudMetadata.blockWithIptables to avoid an ambiguous configuration.

Default: false

singleuser.networkPolicy.egressAllowRules.dnsPortsCloudMetadataServer#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the cloud metadata server via port 53.

Relying on this rule for the singleuser config should go hand in hand with disabling singleuser.cloudMetadata.blockWithIptables to avoid an ambiguous configuration.

Known situations when this rule can be relevant:

  • In GKE clusters with Cloud DNS that is reached at the cloud metadata server’s non-private IP.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

New in version 3.0.0.

Default: true

singleuser.networkPolicy.egressAllowRules.dnsPortsKubeSystemNamespace#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to pods in the kube-system namespace via port 53.

Known situations when this rule can be relevant:

  • GKE, EKS, AKS, and other clusters relying directly on kube-dns or coredns pods in the kube-system namespace.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

New in version 3.0.0.

Default: true

singleuser.networkPolicy.egressAllowRules.dnsPortsPrivateIPs#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to private IPs via port 53.

Known situations when this rule can be relevant:

  • GKE clusters relying on a DNS server indirectly via a a node local DNS cache at an unknown private IP.

Note

This chart doesn’t know how to identify the DNS server that pods will rely on due to variations between how k8s clusters have been setup. Due to that, multiple rules are enabled by default to ensure DNS connectivity.

Warning

This rule is not expected to work in clusters relying on Cilium to enforce the NetworkPolicy rules (includes GKE clusters with Dataplane v2), this is due to a known limitation.

Default: true

singleuser.networkPolicy.egressAllowRules.nonPrivateIPs#

Defaults to true for all network policies.

When enabled this rule allows the respective pod(s) to establish outbound connections to the non-private IP ranges with the exception of the cloud metadata server. This means respective pod(s) can establish connections to the internet but not (say) an unsecured prometheus server running in the same cluster.

Default: true

singleuser.networkPolicy.egressAllowRules.privateIPs#

Defaults to false for singleuser servers, but to true for all other network policies.

Private IPs refer to the IP ranges 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16.

When enabled this rule allows the respective pod(s) to establish outbound connections to the internal k8s cluster. This means users can access the internet but not (say) an unsecured prometheus server running in the same cluster.

Since not all workloads in the k8s cluster may have NetworkPolicies setup to restrict their incoming connections, having this set to false can be a good defense against malicious intent from someone in control of software in these pods.

If possible, try to avoid setting this to true as it gives broad permissions that could be specified more directly via the .egress.

Warning

This rule is not expected to work in clusters relying on Cilium to enforce the NetworkPolicy rules (includes GKE clusters with Dataplane v2), this is due to a known limitation.

Default: false

singleuser.networkPolicy.interNamespaceAccessLabels#

This configuration option determines if both namespaces and pods in other namespaces, that have specific access labels, should be accepted to allow ingress (set to accept), or, if the labels are to be ignored when applied outside the local namespace (set to ignore).

The available access labels for respective NetworkPolicy resources are:

  • hub.jupyter.org/network-access-hub: "true" (hub)

  • hub.jupyter.org/network-access-proxy-http: "true" (proxy.chp, proxy.traefik)

  • hub.jupyter.org/network-access-proxy-api: "true" (proxy.chp)

  • hub.jupyter.org/network-access-singleuser: "true" (singleuser)

Default: "ignore"

singleuser.networkPolicy.allowedIngressPorts#

A rule to allow ingress on these ports will be added no matter what the origin of the request is. The default setting for proxy.chp and proxy.traefik’s networkPolicy configuration is [http, https], while it is [] for other networkPolicies.

Note that these port names or numbers target a Pod’s port name or number, not a k8s Service’s port name or number.

Default: []

singleuser.podNameTemplate#

Passthrough configuration for KubeSpawner.pod_name_template.

singleuser.cpu#

Set CPU limits & guarantees that are enforced for each user.

See the Kubernetes docs for more info.

singleuser.cpu.limit#

singleuser.cpu.guarantee#

singleuser.memory#

Set Memory limits & guarantees that are enforced for each user.

See the Kubernetes docs for more info.

singleuser.memory.limit#

singleuser.memory.guarantee#

Note that this field is referred to as requests by the Kubernetes API.

Default: "1G"

singleuser.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

singleuser.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "quay.io/jupyterhub/k8s-singleuser-sample"

singleuser.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a

singleuser.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

singleuser.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

singleuser.initContainers#

list of initContainers to be run every singleuser pod. See Kubernetes Docs

singleuser:
  initContainers:
    - name: init-myservice
      image: busybox:1.28
      command: ['sh', '-c', 'command1']
    - name: init-mydb
      image: busybox:1.28
      command: ['sh', '-c', 'command2']

Default: []

singleuser.profileList#

For more information about the profile list, see KubeSpawner’s documentation as this is simply a passthrough to that configuration.

Note

The image-pullers are aware of the overrides of images in singleuser.profileList but they won’t be if you configure it in JupyterHub’s configuration of ‘c.KubeSpawner.profile_list.

singleuser:
  profileList:
    - display_name: "Default: Shared, 8 CPU cores"
      description: "Your code will run on a shared machine with CPU only."
      default: True
    - display_name: "Personal, 4 CPU cores & 26GB RAM, 1 NVIDIA Tesla K80 GPU"
      description: "Your code will run a personal machine with a GPU."
      kubespawner_override:
        extra_resource_limits:
          nvidia.com/gpu: "1"

Default: []

singleuser.extraFiles#

A dictionary with extra files to be injected into the pod’s container on startup. This can for example be used to inject: configuration files, custom user interface templates, images, and more.

# NOTE: "hub" is used in this example, but the configuration is the
#       same for "singleuser".
hub:
  extraFiles:
    # The file key is just a reference that doesn't influence the
    # actual file name.
    <file key>:
      # mountPath is required and must be the absolute file path.
      mountPath: <full file path>

      # Choose one out of the three ways to represent the actual file
      # content: data, stringData, or binaryData.
      #
      # data should be set to a mapping (dictionary). It will in the
      # end be rendered to either YAML, JSON, or TOML based on the
      # filename extension that are required to be either .yaml, .yml,
      # .json, or .toml.
      #
      # If your content is YAML, JSON, or TOML, it can make sense to
      # use data to represent it over stringData as data can be merged
      # instead of replaced if set partially from separate Helm
      # configuration files.
      #
      # Both stringData and binaryData should be set to a string
      # representing the content, where binaryData should be the
      # base64 encoding of the actual file content.
      #
      data:
        myConfig:
          myMap:
            number: 123
            string: "hi"
          myList:
            - 1
            - 2
      stringData: |
        hello world!
      binaryData: aGVsbG8gd29ybGQhCg==

      # mode is by default 0644 and you can optionally override it
      # either by octal notation (example: 0400) or decimal notation
      # (example: 256).
      mode: <file system permissions>

Using –set-file

To avoid embedding entire files in the Helm chart configuration, you can use the --set-file flag during helm upgrade to set the stringData or binaryData field.

hub:
  extraFiles:
    my_image:
      mountPath: /usr/local/share/jupyterhub/static/my_image.png

    # Files in /usr/local/etc/jupyterhub/jupyterhub_config.d are
    # automatically loaded in alphabetical order of the final file
    # name when JupyterHub starts.
    my_config:
      mountPath: /usr/local/etc/jupyterhub/jupyterhub_config.d/my_jupyterhub_config.py
# --set-file expects a text based file, so you need to base64 encode
# it manually first.
base64 my_image.png > my_image.png.b64

helm upgrade <...> \
    --set-file hub.extraFiles.my_image.binaryData=./my_image.png.b64 \
    --set-file hub.extraFiles.my_config.stringData=./my_jupyterhub_config.py

Common uses

  1. JupyterHub template customization

    You can replace the default JupyterHub user interface templates in the hub pod by injecting new ones to /usr/local/share/jupyterhub/templates. These can in turn reference custom images injected to /usr/local/share/jupyterhub/static.

  2. JupyterHub standalone file config

    Instead of embedding JupyterHub python configuration as a string within a YAML file through hub.extraConfig, you can inject a standalone .py file into /usr/local/etc/jupyterhub/jupyterhub_config.d that is automatically loaded.

  3. Flexible configuration

    By injecting files, you don’t have to embed them in a docker image that you have to rebuild.

    If your configuration file is a YAML/JSON/TOML file, you can also use data instead of stringData which allow you to set various configuration in separate Helm config files. This can be useful to help dependent charts override only some configuration part of the file, or to allow for the configuration be set through multiple Helm configuration files.

Limitations

  1. File size

    The files in hub.extraFiles and singleuser.extraFiles are respectively stored in their own k8s Secret resource. As k8s Secret’s are limited, typically to 1MB, you will be limited to a total file size of less than 1MB as there is also base64 encoding that takes place reducing available capacity to 75%.

  2. File updates

    The files that are mounted are only set during container startup. This is because we use subPath as is required to avoid replacing the content of the entire directory we mount in.

singleuser.extraEnv#

Extra environment variables that should be set for the user pods.

String literals with $(ENV_VAR_NAME) will be expanded by Kubelet which is a part of Kubernetes. Note that the user pods will already have access to a set of environment variables that you can use, like JUPYTERHUB_USER and JUPYTERHUB_HOST. For more information about these inspect this source code.

singleuser:
  extraEnv:
    # basic notation (for literal values only)
    MY_ENV_VARS_NAME1: "my env var value 1"

    # explicit notation (the "name" field takes precedence)
    USER_NAMESPACE:
      name: USER_NAMESPACE
      valueFrom:
        fieldRef:
          fieldPath: metadata.namespace

    # implicit notation (the "name" field is implied)
    PREFIXED_USER_NAMESPACE:
      value: "my-prefix-$(USER_NAMESPACE)"
    SECRET_VALUE:
      valueFrom:
        secretKeyRef:
          name: my-k8s-secret
          key: password

For more information, see the Kubernetes EnvVar specification.

singleuser.nodeSelector#

An object with key value pairs representing labels. K8s Nodes are required to have match all these labels for this Pod to scheduled on them.

disktype: ssd
nodetype: awesome

See the Kubernetes documentation for more details.

singleuser.extraTolerations#

Tolerations allow a pod to be scheduled on nodes with taints. These tolerations are additional tolerations to the tolerations common to all pods of a their respective kind (scheduling.corePods.tolerations, scheduling.userPods.tolerations).

Pass this field an array of Toleration objects.

See the Kubernetes docs for more info.

Default: []

singleuser.extraNodeAffinity#

Affinities describe where pods prefer or require to be scheduled, they may prefer or require a node where they are to be scheduled to have a certain label (node affinity). They may also require to be scheduled in proximity or with a lack of proximity to another pod (pod affinity and anti pod affinity).

See the Kubernetes docs for more info.

singleuser.extraNodeAffinity.required#

Pass this field an array of NodeSelectorTerm objects.

Default: []

singleuser.extraNodeAffinity.preferred#

Pass this field an array of PreferredSchedulingTerm objects.

Default: []

singleuser.extraPodAffinity#

See the description of singleuser.extraNodeAffinity.

singleuser.extraPodAffinity.required#

Pass this field an array of PodAffinityTerm objects.

Default: []

singleuser.extraPodAffinity.preferred#

Pass this field an array of WeightedPodAffinityTerm objects.

Default: []

singleuser.extraPodAntiAffinity#

See the description of singleuser.extraNodeAffinity.

singleuser.extraPodAntiAffinity.required#

Pass this field an array of PodAffinityTerm objects.

Default: []

singleuser.extraPodAntiAffinity.preferred#

Pass this field an array of WeightedPodAffinityTerm objects.

Default: []

singleuser.cloudMetadata#

Please refer to dedicated section in the Helm chart documentation for more information about this.

singleuser.cloudMetadata.blockWithIptables#

Default: true

singleuser.cloudMetadata.ip#

Default: "169.254.169.254"

singleuser.cmd#

Passthrough configuration for KubeSpawner.cmd. The default is “jupyterhub-singleuser”. Use cmd: null to launch a custom CMD from the image, which must launch jupyterhub-singleuser or an equivalent process eventually. For example: Jupyter’s docker-stacks images.

Default: "jupyterhub-singleuser"

singleuser.defaultUrl#

Passthrough configuration for KubeSpawner.default_url.

singleuser.events#

Passthrough configuration for KubeSpawner.events_enabled.

Default: true

singleuser.extraAnnotations#

Passthrough configuration for KubeSpawner.extra_annotations.

singleuser.extraContainers#

Passthrough configuration for KubeSpawner.extra_containers.

Default: []

singleuser.extraLabels#

Passthrough configuration for KubeSpawner.extra_labels.

singleuser.extraPodConfig#

Passthrough configuration for KubeSpawner.extra_pod_config.

singleuser.extraResource#

singleuser.extraResource.guarantees#

Passthrough configuration for KubeSpawner.extra_resource_guarantees.

singleuser.extraResource.limits#

Passthrough configuration for KubeSpawner.extra_resource_limits.

singleuser.fsGid#

Passthrough configuration for KubeSpawner.fs_gid.

Default: 100

singleuser.lifecycleHooks#

Passthrough configuration for KubeSpawner.lifecycle_hooks.

singleuser.lifecycleHooks.postStart#

singleuser.lifecycleHooks.preStop#

singleuser.networkTools#

This configuration section refers to configuration of a conditionally created initContainer for the user pods with a purpose to block a specific IP address.

This initContainer will be created if singleuser.cloudMetadata.blockWithIptables is set to true.

singleuser.networkTools.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

singleuser.networkTools.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "quay.io/jupyterhub/k8s-network-tools"

singleuser.networkTools.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a
singleuser.networkTools.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

singleuser.networkTools.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

singleuser.networkTools.resources#

A k8s native specification of resources, see the documentation.

singleuser.serviceAccountName#

Passthrough configuration for KubeSpawner.service_account.

singleuser.startTimeout#

Passthrough configuration for KubeSpawner.start_timeout.

Default: 300

singleuser.storage#

This section configures KubeSpawner directly to some extent but also indirectly through Helm chart specific configuration options such as singleuser.storage.type.

singleuser.storage.capacity#

Configures KubeSpawner.storage_capacity.

See the KubeSpawner documentation for more information.

Default: "10Gi"

singleuser.storage.dynamic#

singleuser.storage.dynamic.pvcNameTemplate#

Configures KubeSpawner.pvc_name_template which will be the resource name of the PVC created by KubeSpawner for each user if needed.

Default: "claim-{username}{servername}"

singleuser.storage.dynamic.storageAccessModes#

Configures KubeSpawner.storage_access_modes.

See KubeSpawners documentation and the k8s documentation for more information.

Default: ["ReadWriteOnce"]

singleuser.storage.dynamic.storageClass#

Configures KubeSpawner.storage_class, which can be an explicit StorageClass to dynamically provision storage for the PVC that KubeSpawner will create.

There is of a default StorageClass available in k8s clusters for use if this is unspecified.

singleuser.storage.dynamic.volumeNameTemplate#

Configures KubeSpawner.volume_name_template, which is the name to reference from the containers volumeMounts section.

Default: "volume-{username}{servername}"

singleuser.storage.extraLabels#

Configures KubeSpawner.storage_extra_labels. Note that these labels are set on the PVC during creation only and won’t be updated after creation.

singleuser.storage.extraVolumeMounts#

Additional volume mounts for the Container. Use a k8s native syntax.

Default: []

singleuser.storage.extraVolumes#

Additional volumes for the Pod. Use a k8s native syntax.

Default: []

singleuser.storage.homeMountPath#

The location within the container where the home folder storage should be mounted.

Default: "/home/jovyan"

singleuser.storage.static#

singleuser.storage.static.pvcName#

Configures KubeSpawner.pvc_claim_name to reference pre-existing storage.

singleuser.storage.static.subPath#

Configures the subPath field of a KubeSpawner.volume_mounts entry added by the Helm chart.

Path within the volume from which the container’s volume should be mounted.

Default: "{username}"

singleuser.storage.type#

Decide if you want storage to be provisioned dynamically (dynamic), or if you want to attach existing storage (static), or don’t want any storage to be attached (none).

Default: "dynamic"

singleuser.allowPrivilegeEscalation#

Passthrough configuration for KubeSpawner.allow_privilege_escalation.

Default: false

singleuser.uid#

Passthrough configuration for KubeSpawner.uid.

This dictates as what user the main container will start up as.

As an example of when this is needed, consider if you want to enable sudo rights for some of your users. This can be done by starting up as root, enabling it from the container in a startup script, and then transitioning to the normal user.

Default is 1000, set to null to use the container’s default.

Default: 1000

scheduling#

Objects for customizing the scheduling of various pods on the nodes and related labels.

scheduling.userScheduler#

The user scheduler is making sure that user pods are scheduled tight on nodes, this is useful for autoscaling of user node pools.

scheduling.userScheduler.enabled#

Enables the user scheduler.

Default: true

scheduling.userScheduler.revisionHistoryLimit#

Configures the resource’s spec.revisionHistoryLimit. This is available for Deployment, StatefulSet, and DaemonSet resources.

See the Kubernetes docs for more info.

scheduling.userScheduler.replicas#

You can have multiple schedulers to share the workload or improve availability on node failure.

Default: 2

scheduling.userScheduler.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

scheduling.userScheduler.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "registry.k8s.io/kube-scheduler"

scheduling.userScheduler.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a

Default: "v1.28.8"

scheduling.userScheduler.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

scheduling.userScheduler.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

scheduling.userScheduler.pdb#

Configure a PodDisruptionBudget for this Deployment.

These are disabled by default for our deployments that don’t support being run in parallel with multiple replicas. Only the user-scheduler currently supports being run in parallel with multiple replicas. If they are enabled for a Deployment with only one replica, they will block kubectl drain of a node for example.

Note that if you aim to block scaling down a node with the hub/proxy/autohttps pod that would cause disruptions of the deployment, then you should instead annotate the pods of the Deployment as described here.

"cluster-autoscaler.kubernetes.io/safe-to-evict": "false"

See the Kubernetes documentation for more details about disruptions.

scheduling.userScheduler.pdb.enabled#

Decides if a PodDisruptionBudget is created targeting the Deployment’s pods.

Default: true

scheduling.userScheduler.pdb.maxUnavailable#

The maximum number of pods that can be unavailable during voluntary disruptions.

Default: 1

scheduling.userScheduler.pdb.minAvailable#

The minimum number of pods required to be available during voluntary disruptions.

scheduling.userScheduler.nodeSelector#

An object with key value pairs representing labels. K8s Nodes are required to have match all these labels for this Pod to scheduled on them.

disktype: ssd
nodetype: awesome

See the Kubernetes documentation for more details.

scheduling.userScheduler.tolerations#

Tolerations allow a pod to be scheduled on nodes with taints. These tolerations are additional tolerations to the tolerations common to all pods of a their respective kind (scheduling.corePods.tolerations, scheduling.userPods.tolerations).

Pass this field an array of Toleration objects.

See the Kubernetes docs for more info.

Default: []

scheduling.userScheduler.labels#

Extra labels to add to the userScheduler pods.

See the Kubernetes docs to learn more about labels.

scheduling.userScheduler.annotations#

Extra annotations to add to the user-scheduler pods.

scheduling.userScheduler.containerSecurityContext#

A k8s native specification of the container’s security context, see the documentation for details.

scheduling.userScheduler.logLevel#

Corresponds to the verbosity level of logging made by the kube-scheduler binary running within the user-scheduler pod.

Default: 4

scheduling.userScheduler.plugins#

These plugins refers to kube-scheduler plugins as documented here.

The user-scheduler is really just a kube-scheduler configured in a way to pack users tight on nodes using these plugins. See values.yaml for information about the default plugins.

scheduling.userScheduler.pluginConfig#

Individually activated plugins can be configured further.

Default: [{"name": "NodeResourcesFit", "args": {"scoringStrategy": {"resources": [{"name": "cpu", "weight": 1}, {"name": "memory", "weight": 1}], "type": "MostAllocated"}}}]

scheduling.userScheduler.resources#

A k8s native specification of resources, see the documentation.

scheduling.userScheduler.serviceAccount#

Configuration for a k8s ServiceAccount dedicated for use by the specific pod which this configuration is nested under.

scheduling.userScheduler.serviceAccount.create#

Whether or not to create the ServiceAccount resource.

Default: true

scheduling.userScheduler.serviceAccount.name#

This configuration serves multiple purposes:

  • It will be the serviceAccountName referenced by related Pods.

  • If create is set, the created ServiceAccount resource will be named like this.

  • If rbac.create is set, the associated (Cluster)RoleBindings will bind to this name.

If not explicitly provided, a default name will be used.

scheduling.userScheduler.serviceAccount.annotations#

Kubernetes annotations to apply to the k8s ServiceAccount.

scheduling.userScheduler.extraPodSpec#

Arbitrary extra k8s pod specification as a YAML object. The default value of this setting is an empty object, i.e. no extra configuration. The value of this property is augmented to the pod specification as-is.

This is a powerful tool for expert k8s administrators with advanced configuration requirements. This setting should only be used for configuration that cannot be accomplished through the other settings. Misusing this setting can break your deployment and/or compromise your system security.

This is one of four related settings for inserting arbitrary pod specification:

  1. hub.extraPodSpec

  2. proxy.chp.extraPodSpec

  3. proxy.traefik.extraPodSpec

  4. scheduling.userScheduler.extraPodSpec

One real-world use of these settings is to enable host networking. For example, to configure host networking for the hub pod, add the following to your helm configuration values:

hub:
  extraPodSpec:
    hostNetwork: true
    dnsPolicy: ClusterFirstWithHostNet

Likewise, to configure host networking for the proxy pod, add the following:

proxy:
  chp:
    extraPodSpec:
      hostNetwork: true
      dnsPolicy: ClusterFirstWithHostNet

N.B. Host networking has special security implications and can easily break your deployment. This is an example—not an endorsement.

See PodSpec for the latest pod resource specification.

scheduling.podPriority#

Pod Priority is used to allow real users evict user placeholder pods that in turn by entering a Pending state can trigger a scale up by a cluster autoscaler.

Having this option enabled only make sense if the following conditions are met:

  1. A cluster autoscaler is installed.

  2. user-placeholer pods are configured to have a priority equal or higher than the cluster autoscaler’s “priority cutoff” so that the cluster autoscaler scales up a node in advance for a pending user placeholder pod.

  3. Normal user pods have a higher priority than the user-placeholder pods.

  4. Image puller pods have a priority between normal user pods and user-placeholder pods.

Note that if the default priority cutoff if not configured on cluster autoscaler, it will currently default to 0, and that in the future this is meant to be lowered. If your cloud provider is installing the cluster autoscaler for you, they may also configure this specifically.

Recommended settings for a cluster autoscaler…

… with a priority cutoff of -10 (GKE):

podPriority:
  enabled: true
  globalDefault: false
  defaultPriority: 0
  imagePullerPriority: -5
  userPlaceholderPriority: -10

… with a priority cutoff of 0:

podPriority:
  enabled: true
  globalDefault: true
  defaultPriority: 10
  imagePullerPriority: 5
  userPlaceholderPriority: 0

scheduling.podPriority.enabled#

Default: false

scheduling.podPriority.globalDefault#

Warning! This will influence all pods in the cluster.

The priority a pod usually get is 0. But this can be overridden with a PriorityClass resource if it is declared to be the global default. This configuration option allows for the creation of such global default.

Default: false

scheduling.podPriority.defaultPriority#

The actual value for the default pod priority.

Default: 0

scheduling.podPriority.imagePullerPriority#

The actual value for the [hook|continuous]-image-puller pods’ priority.

Default: -5

scheduling.podPriority.userPlaceholderPriority#

The actual value for the user-placeholder pods’ priority.

Default: -10

scheduling.userPlaceholder#

User placeholders simulate users but will thanks to PodPriority be evicted by the cluster autoscaler if a real user shows up. In this way placeholders allow you to create a headroom for the real users and reduce the risk of a user having to wait for a node to be added. Be sure to use the the continuous image puller as well along with placeholders, so the images are also available when real users arrive.

To test your setup efficiently, you can adjust the amount of user placeholders with the following command:

# Configure to have 3 user placeholders
kubectl scale sts/user-placeholder --replicas=3

scheduling.userPlaceholder.enabled#

Default: true

scheduling.userPlaceholder.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

scheduling.userPlaceholder.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "registry.k8s.io/pause"

scheduling.userPlaceholder.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a

Default: "3.9"

scheduling.userPlaceholder.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

scheduling.userPlaceholder.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

scheduling.userPlaceholder.revisionHistoryLimit#

Configures the resource’s spec.revisionHistoryLimit. This is available for Deployment, StatefulSet, and DaemonSet resources.

See the Kubernetes docs for more info.

scheduling.userPlaceholder.replicas#

How many placeholder pods would you like to have?

Default: 0

scheduling.userPlaceholder.labels#

Extra labels to add to the userPlaceholder pods.

See the Kubernetes docs to learn more about labels.

scheduling.userPlaceholder.annotations#

Extra annotations to add to the placeholder pods.

scheduling.userPlaceholder.resources#

Unless specified here, the placeholder pods will request the same resources specified for the real singleuser pods.

scheduling.userPlaceholder.containerSecurityContext#

A k8s native specification of the container’s security context, see the documentation for details.

scheduling.corePods#

These settings influence the core pods like the hub, proxy and user-scheduler pods. These settings influence all pods considered core pods, namely:

  • hub

  • proxy

  • autohttps

  • hook-image-awaiter

  • user-scheduler

By defaults, the tolerations are:

  • hub.jupyter.org/dedicated=core:NoSchedule

  • hub.jupyter.org_dedicated=core:NoSchedule

Note that tolerations set here are combined with the respective components dedicated tolerations, and that _ is available in case / isn’t allowed in the clouds tolerations.

scheduling.corePods.tolerations#

Tolerations allow a pod to be scheduled on nodes with taints. These tolerations are additional tolerations to the tolerations common to all pods of a their respective kind (scheduling.corePods.tolerations, scheduling.userPods.tolerations).

Pass this field an array of Toleration objects.

See the Kubernetes docs for more info.

Default: [{"key": "hub.jupyter.org/dedicated", "operator": "Equal", "value": "core", "effect": "NoSchedule"}, {"key": "hub.jupyter.org_dedicated", "operator": "Equal", "value": "core", "effect": "NoSchedule"}]

scheduling.corePods.nodeAffinity#

Where should pods be scheduled? Perhaps on nodes with a certain label is preferred or even required?

scheduling.corePods.nodeAffinity.matchNodePurpose#

Decide if core pods ignore, prefer or require to schedule on nodes with this label:

hub.jupyter.org/node-purpose=core

Default: "prefer"

scheduling.userPods#

These settings influence all pods considered user pods, namely:

  • user-placeholder

  • hook-image-puller

  • continuous-image-puller

  • jupyter-

By defaults, the tolerations are:

  • hub.jupyter.org/dedicated=core:NoSchedule

  • hub.jupyter.org_dedicated=core:NoSchedule

Note that tolerations set here are combined with the respective components dedicated tolerations, and that _ is available in case / isn’t allowed in the clouds tolerations.

scheduling.userPods.tolerations#

Tolerations allow a pod to be scheduled on nodes with taints. These tolerations are additional tolerations to the tolerations common to all pods of a their respective kind (scheduling.corePods.tolerations, scheduling.userPods.tolerations).

Pass this field an array of Toleration objects.

See the Kubernetes docs for more info.

Default: [{"key": "hub.jupyter.org/dedicated", "operator": "Equal", "value": "user", "effect": "NoSchedule"}, {"key": "hub.jupyter.org_dedicated", "operator": "Equal", "value": "user", "effect": "NoSchedule"}]

scheduling.userPods.nodeAffinity#

Where should pods be scheduled? Perhaps on nodes with a certain label is preferred or even required?

scheduling.userPods.nodeAffinity.matchNodePurpose#

Decide if user pods ignore, prefer or require to schedule on nodes with this label:

hub.jupyter.org/node-purpose=user

Default: "prefer"

ingress#

ingress.enabled#

Enable the creation of a Kubernetes Ingress to proxy-public service.

See Advanced Topics — Zero to JupyterHub with Kubernetes 0.7.0 documentation for more details.

Default: false

ingress.annotations#

Annotations to apply to the Ingress resource.

See the Kubernetes documentation for more details about annotations.

ingress.ingressClassName#

Maps directly to the Ingress resource’s `spec.ingressClassName``.

See the Kubernetes documentation for more details.

ingress.hosts#

List of hosts to route requests to the proxy.

Default: []

ingress.pathSuffix#

Suffix added to Ingress’s routing path pattern.

Specify * if your ingress matches path by glob pattern.

ingress.pathType#

The path type to use. The default value is ‘Prefix’.

See the Kubernetes documentation for more details about path types.

Default: "Prefix"

ingress.tls#

TLS configurations for Ingress.

See the Kubernetes documentation for more details about annotations.

Default: []

prePuller#

prePuller.revisionHistoryLimit#

Configures the resource’s spec.revisionHistoryLimit. This is available for Deployment, StatefulSet, and DaemonSet resources.

See the Kubernetes docs for more info.

prePuller.labels#

Extra labels to add to the pre puller job pods.

See the Kubernetes docs to learn more about labels.

prePuller.annotations#

Annotations to apply to the hook and continous image puller pods. One example use case is to disable istio sidecars which could interfere with the image pulling.

prePuller.resources#

These are standard Kubernetes resources with requests and limits for cpu and memory. They will be used on the containers in the pods pulling images. These should be set extremely low as the containers shut down directly or is a pause container that just idles.

They were made configurable as usage of ResourceQuota may require containers in the namespace to have explicit resources set.

prePuller.extraTolerations#

Tolerations allow a pod to be scheduled on nodes with taints. These tolerations are additional tolerations to the tolerations common to all pods of a their respective kind (scheduling.corePods.tolerations, scheduling.userPods.tolerations).

Pass this field an array of Toleration objects.

See the Kubernetes docs for more info.

Default: []

prePuller.hook#

See the optimization section for more details.

prePuller.hook.enabled#

Default: true

prePuller.hook.pullOnlyOnChanges#

Pull only if changes have been made to the images to pull, or more accurately if the hook-image-puller daemonset has changed in any way.

Default: true

prePuller.hook.podSchedulingWaitDuration#

The hook-image-awaiter has a criteria to await all the hook-image-puller DaemonSet’s pods to both schedule and finish their image pulling. This flag can be used to relax this criteria to instead only await the pods that has already scheduled to finish image pulling after a certain duration.

The value of this is that sometimes the newly created hook-image-puller pods cannot be scheduled because nodes are full, and then it probably won’t make sense to block a helm upgrade.

An infinite duration to wait for pods to schedule can be represented by -1. This was the default behavior of version 0.9.0 and earlier.

Default: 10

prePuller.hook.nodeSelector#

An object with key value pairs representing labels. K8s Nodes are required to have match all these labels for this Pod to scheduled on them.

disktype: ssd
nodetype: awesome

See the Kubernetes documentation for more details.

prePuller.hook.tolerations#

Tolerations allow a pod to be scheduled on nodes with taints. These tolerations are additional tolerations to the tolerations common to all pods of a their respective kind (scheduling.corePods.tolerations, scheduling.userPods.tolerations).

Pass this field an array of Toleration objects.

See the Kubernetes docs for more info.

Default: []

prePuller.hook.containerSecurityContext#

A k8s native specification of the container’s security context, see the documentation for details.

prePuller.hook.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

prePuller.hook.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "quay.io/jupyterhub/k8s-image-awaiter"

prePuller.hook.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a
prePuller.hook.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

prePuller.hook.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

prePuller.hook.resources#

A k8s native specification of resources, see the documentation.

prePuller.hook.serviceAccount#

Configuration for a k8s ServiceAccount dedicated for use by the specific pod which this configuration is nested under.

prePuller.hook.serviceAccount.create#

Whether or not to create the ServiceAccount resource.

Default: true

prePuller.hook.serviceAccount.name#

This configuration serves multiple purposes:

  • It will be the serviceAccountName referenced by related Pods.

  • If create is set, the created ServiceAccount resource will be named like this.

  • If rbac.create is set, the associated (Cluster)RoleBindings will bind to this name.

If not explicitly provided, a default name will be used.

prePuller.hook.serviceAccount.annotations#

Kubernetes annotations to apply to the k8s ServiceAccount.

prePuller.continuous#

See the optimization section for more details.

Note

If used with a Cluster Autoscaler (an autoscaling node pool), also add user-placeholders and enable pod priority.

prePuller.continuous.enabled#

Default: true

prePuller.pullProfileListImages#

The singleuser.profileList configuration can provide a selection of images. This option determines if all images identified there should be pulled, both by the hook and continuous pullers.

Images are looked for under kubespawner_override, and also profile_options.choices.kubespawner_override since version 3.2.0.

The reason to disable this, is that if you have for example 10 images which start pulling in order from 1 to 10, a user that arrives and wants to start a pod with image number 10 will need to wait for all images to be pulled, and then it may be preferable to just let the user arriving wait for a single image to be pulled on arrival.

Default: true

prePuller.extraImages#

See the optimization section for more details.

prePuller:
  extraImages:
    my-extra-image-i-want-pulled:
      name: jupyter/all-spark-notebook
      tag: 2343e33dec46

prePuller.containerSecurityContext#

A k8s native specification of the container’s security context, see the documentation for details.

prePuller.pause#

The image-puller pods rely on initContainer to pull all images, and their actual container when they are done is just running a pause container. These are settings for that pause container.

prePuller.pause.containerSecurityContext#

A k8s native specification of the container’s security context, see the documentation for details.

prePuller.pause.image#

Set custom image name, tag, pullPolicy, or pullSecrets for the pod.

prePuller.pause.image.name#

The name of the image, without the tag.

# example name
gcr.io/my-project/my-image

Default: "registry.k8s.io/pause"

prePuller.pause.image.tag#

The tag of the image to pull. This is the value following : in complete image specifications.

# example tags
v1.11.1
zhy270a

Default: "3.9"

prePuller.pause.image.pullPolicy#

Configures the Pod’s spec.imagePullPolicy.

See the Kubernetes docs for more info.

prePuller.pause.image.pullSecrets#

A list of references to existing Kubernetes Secrets with credentials to pull the image.

This Pod’s final imagePullSecrets k8s specification will be a combination of:

  1. This list of k8s Secrets, specific for this pod.

  2. The list of k8s Secrets, for use by all pods in the Helm chart, declared in this Helm charts configuration called imagePullSecrets.

  3. A k8s Secret, for use by all pods in the Helm chart, if conditionally created from image registry credentials provided under imagePullSecret if imagePullSecret.create is set to true.

# example - k8s native syntax
pullSecrets:
  - name: my-k8s-secret-with-image-registry-credentials

# example - simplified syntax
pullSecrets:
  - my-k8s-secret-with-image-registry-credentials

Default: []

custom#

Additional values to pass to the Hub. JupyterHub will not itself look at these, but you can read values in your own custom config via hub.extraConfig. For example:

custom:
  myHost: "https://example.horse"
hub:
  extraConfig:
    myConfig.py: |
      c.MyAuthenticator.host = get_config("custom.myHost")

cull#

The jupyterhub-idle-culler can run as a JupyterHub managed service to cull running servers.

cull.enabled#

Enable/disable use of jupyter-idle-culler.

Default: true

cull.users#

See the --cull-users flag. Default: false

cull.adminUsers#

See the --cull-admin-users flag. Default: true

cull.removeNamedServers#

See the --remove-named-servers flag. Default: false

cull.timeout#

See the --timeout flag. Default: 3600

cull.every#

See the --cull-every flag. Default: 600

cull.concurrency#

See the --concurrency flag. Default: 10

cull.maxAge#

See the --max-age flag. Default: 0

debug#

debug.enabled#

Increases the loglevel throughout the resources in the Helm chart.

Default: false

rbac#

rbac.enabled#

Note

Removed in version 2.0.0. If you have been using rbac.enable=false (strongly discouraged), then the equivalent configuration would be:

rbac:
  create: false
hub:
  serviceAccount:
    create: false
proxy:
  traefik:
    serviceAccount:
      create: false
scheduling:
  userScheduler:
    serviceAccount:
      create: false
prePuller:
  hook:
    serviceAccount:
      create: false

rbac.create#

Decides if (Cluster)Role and (Cluster)RoleBinding resources are created and bound to the configured serviceAccounts.

Default: true

global#

global.safeToShowValues#

A flag that should only be set to true temporarily when experiencing a deprecation message that contain censored content that you wish to reveal.

Default: false