In this article, we'll guide how to set up Kafka-UI with role-based access control.
First of all, you'd need to set up authentication method(s). Refer to this article for OAuth2 setup.
First of all, you have to decide if either:
- You wish to store all roles in a separate config file
- Or within a main config file
This is how you include one more file to start with a docker-compose example:
services:
kafka-ui:
container_name: kafka-ui
image: provectuslabs/kafka-ui:latest
environment:
KAFKA_CLUSTERS_0_NAME: local
# other properties, omitted
SPRING_CONFIG_ADDITIONAL-LOCATION: /roles.yml
volumes:
- /tmp/roles.yml:/roles.yml
Alternatively, you can append the roles file contents to your main config file.
In the roles file we define roles, duh. Every each role has access to defined clusters:
rbac:
roles:
- name: "memelords"
clusters:
- local
- dev
- staging
- prod
A role also has a list of subjects which are the entities we will use to assign roles to. They are provider-dependant, in general, they can be users, groups, or some other entities (github orgs, google domains, LDAP queries, etc.) In this example we define a role memelords
that will contain all the users within the Google domain memelord.lol
and, additionally, a GitHub user Haarolean
. You can combine as many subjects as you want within a role.
- name: "memelords"
subjects:
- provider: oauth_google
type: domain
value: "memelord.lol"
- provider: oauth_github
type: user
value: "Haarolean"
A list of supported providers and corresponding subject fetch mechanism:
- oauth_google:
user
,domain
- oauth_github:
user
,organization
- oauth_cognito:
user
,group
- ldap:
group
- ldap_ad: (unsupported yet, will do in 0.8 release)
Find the more detailed examples in a full example file lower.
The next thing which is present in your roles file is, surprisingly, permissions. They consist of:
- Resource Can be one of the:
CLUSTERCONFIG
,TOPIC
,CONSUMER
,SCHEMA
,CONNECT
,KSQL
,ACL
. - The resource value is either a fixed string or a regular expression identifying a resource. Value is not applicable to
clusterconfig
andksql
resources. Please do not fill it out. - Actions It's a list of actions (the possible values depend on the resource, see the lists below) that will be applied to the certain permission. Also, note, there's a special action for any of the resources called "all", it will virtually grant all the actions within the corresponding resource. An example for enabling viewing and creating topics whose name start with "derp":
permissions:
- resource: topic
value: "derp.*"
actions: [ VIEW, CREATE ]
Actions
A list of all the actions for the corresponding resources (please note neither resource nor action names are case-sensitive):
applicationconfig
:view
,edit
clusterconfig
:view
,edit
topic
:view
,create
,edit
,delete
,messages_read
,messages_produce
,messages_delete
consumer
:view
,delete
,reset_offsets
schema
:view
,create
,delete
,edit
,modify_global_compatibility
connect
:view
,edit
,create
,restart
ksql
:execute
acl
:view
,edit
A complete file example:
rbac:
roles:
- name: "memelords"
clusters:
- local
- dev
- staging
- prod
subjects:
- provider: oauth_google
type: domain
value: "memelord.lol"
- provider: oauth_google
type: user
value: "kek@memelord.lol"
- provider: oauth_github
type: organization
value: "memelords_team"
- provider: oauth_github
type: user
value: "memelord"
- provider: oauth_cognito
type: user
value: "username"
- provider: oauth_cognito
type: group
value: "memelords"
- provider: ldap
type: group
value: "admin_staff"
- provider: ldap_ad # NOT YET SUPPORTED, SEE ISSUE 3741
type: user
value: "cn=germanosin,dc=planetexpress,dc=com"
permissions:
- resource: applicationconfig
# value not applicable for applicationconfig
actions: [ "view", "edit" ] # can be with or without quotes
- resource: clusterconfig
# value not applicable for clusterconfig
actions: [ "view", "edit" ]
- resource: topic
value: "ololo.*"
actions: # can be a multiline list
- VIEW # can be upper or lowercase
- CREATE
- EDIT
- DELETE
- MESSAGES_READ
- MESSAGES_PRODUCE
- MESSAGES_DELETE
- resource: consumer
value: "\_confluent-ksql.*"
actions: [ VIEW, DELETE, RESET_OFFSETS ]
- resource: schema
value: "blah.*"
actions: [ VIEW, CREATE, DELETE, EDIT, MODIFY_GLOBAL_COMPATIBILITY ]
- resource: connect
value: "local"
actions: [ view, edit, create ]
# connectors selector not implemented yet, use connects
# selector:
# connector:
# name: ".*"
# class: 'com.provectus.connectorName'
- resource: ksql
# value not applicable for ksql
actions: [ execute ]
- resource: acl
# value not applicable for acl
actions: [ view, edit ]
A read-only setup:
rbac:
roles:
- name: "readonly"
clusters:
# FILL THIS
subjects:
# FILL THIS
permissions:
- resource: clusterconfig
actions: [ "view" ]
- resource: topic
value: ".*"
actions:
- VIEW
- MESSAGES_READ
- resource: consumer
value: ".*"
actions: [ view ]
- resource: schema
value: ".*"
actions: [ view ]
- resource: connect
value: ".*"
actions: [ view ]
- resource: acl
actions: [ view ]
An admin-group setup example:
rbac:
roles:
- name: "admins"
clusters:
# FILL THIS
subjects:
# FILL THIS
permissions:
- resource: applicationconfig
actions: all
- resource: clusterconfig
actions: all
- resource: topic
value: ".*"
actions: all
- resource: consumer
value: ".*"
actions: all
- resource: schema
value: ".*"
actions: all
- resource: connect
value: ".*"
actions: all
- resource: ksql
actions: all
- resource: acl
actions: [ view ]