mariadb-operator/mariadb-operator

[Bug] Operator fails to update corresponding MaxScale object when modifying maxscale module parameters in MariaDB

Closed this issue · 6 comments

luohaha3123 commented

Documentation

Describe the bug
I encountered an issue where the operator fails to update the corresponding MaxScale object when I modify the maxscale module parameters in MariaDB. Specifically, the switchover_on_low_disk_space field is not being updated correctly.

The MariaDB object is:

apiVersion: k8s.mariadb.com/v1alpha1
kind: MariaDB
metadata:
  name: mariadb-test
  namespace: kube-system
spec:
  affinity:
    enableAntiAffinity: true
    podAntiAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
      - labelSelector:
          matchExpressions:
          - key: app.kubernetes.io/instance
            operator: In
            values:
            - mariadb-test
        topologyKey: kubernetes.io/hostname
  database: test
  image: mariadb:10.8.2
  maxScale:
    auth:
      adminPasswordSecretKeyRef:
        key: password
        name: my-test
      adminUsername: mariadb-operator
      clientPasswordSecretKeyRef:
        key: password
        name: my-test
      clientUsername: mariadb-test-maxscale-client
      generate: true
      monitorPasswordSecretKeyRef:
        key: password
        name: my-test
      monitorUsername: mariadb-test-maxscale-monitor
      serverPasswordSecretKeyRef:
        key: password
        name: my-test
      serverUsername: mariadb-test-maxscale-server
      syncPasswordSecretKeyRef:
        key: password
        name: my-test
      syncUsername: mariadb-test-maxscale-sync
    config:
      sync:
        database: mysql
        interval: 5s
        timeout: 15s
      volumeClaimTemplate:
        accessModes:
        - ReadWriteOnce
        resources:
          requests:
            storage: 100Mi
        storageClassName: yj-silver24
    connection:
      port: 3306
      secretName: mxs-test-conn
    enabled: true
    kubernetesService:
      type: NodePort
    monitor:
      cooperativeMonitoring: majority_of_all
      interval: 2s
      module: ""
      name: ""
      params:
        auto_failover: "true"
        auto_rejoin: "true"
        switchover_on_low_disk_space: "false"
    replicas: 2
    services:
    - listener:
        name: rw-listener
        params:
          connection_metadata: tx_isolation=auto
        port: 3306
        protocol: MariaDBProtocol
      name: rw-router
      params:
        master_accept_reads: "true"
        max_replication_lag: 3s
        max_slave_connections: "255"
        transaction_replay: "true"
        transaction_replay_attempts: "10"
        transaction_replay_timeout: 5s
      router: readwritesplit
    - listener:
        name: ""
        port: 3307
      name: rconn-master-router
      params:
        master_accept_reads: "true"
        max_replication_lag: 3s
        router_options: master
      router: readconnroute
    - listener:
        name: ""
        port: 3308
      name: rconn-slave-router
      params:
        max_replication_lag: 3s
        router_options: slave
      router: readconnroute
  maxScaleRef:
    name: mariadb-test-maxscale
    namespace: kube-system
  passwordSecretKeyRef:
    key: root-password
    name: mariadb
  port: 3306
  primaryService:
    type: ClusterIP
  replicas: 3
  replication:
    enabled: true
    primary:
      automaticFailover: true
      podIndex: 0
    probesEnabled: false
    replica:
      connectionRetries: 10
      connectionTimeout: 10s
      gtid: CurrentPos
      replPasswordSecretKeyRef:
        key: password
        name: mariadb
      syncTimeout: 10s
      waitPoint: AfterSync
    syncBinlog: true
  rootEmptyPassword: false
  rootPasswordSecretKeyRef:
    key: root-password
    name: mariadb
  secondaryService:
    type: ClusterIP
  service:
    type: ClusterIP
  serviceAccountName: mariadb-test
  storage:
    ephemeral: false
    resizeInUseVolumes: true
    size: 3Gi
    storageClassName: yj-silver24
    volumeClaimTemplate:
      accessModes:
      - ReadWriteOnce
      resources:
        requests:
          storage: 3Gi
      storageClassName: yj-silver24
    waitForVolumeResize: true
  tolerations:
  - effect: NoSchedule
    key: k8s.mariadb.com/ha
    operator: Exists
  username: test

The MaxScale object is:

apiVersion: k8s.mariadb.com/v1alpha1
kind: MaxScale
metadata:
  name: mariadb-test-maxscale
spec:
  admin:
    guiEnabled: true
    port: 8989
  auth:
    adminPasswordSecretKeyRef:
      key: password
      name: my-test
    adminUsername: mariadb-operator
    clientMaxConnections: 60
    clientPasswordSecretKeyRef:
      key: password
      name: my-test
    clientUsername: mariadb-test-maxscale-client
    deleteDefaultAdmin: true
    generate: true
    monitorMaxConnections: 60
    monitorPasswordSecretKeyRef:
      key: password
      name: my-test
    monitorUsername: mariadb-test-maxscale-monitor
    serverMaxConnections: 60
    serverPasswordSecretKeyRef:
      key: password
      name: my-test
    serverUsername: mariadb-test-maxscale-server
    syncMaxConnections: 60
    syncPasswordSecretKeyRef:
      key: password
      name: my-test
    syncUsername: mariadb-test-maxscale-sync
  config:
    sync:
      database: mysql
      interval: 5s
      timeout: 10s
    volumeClaimTemplate:
      accessModes:
      - ReadWriteOnce
      resources:
        requests:
          storage: 100Mi
      storageClassName: yj-silver24
  connection:
    port: 3306
    secretName: mxs-test-conn
  image: mariadb/maxscale:23.08
  kubernetesService:
    type: NodePort
  mariaDbRef:
    name: mariadb-test
    namespace: kube-system
    waitForIt: false
  monitor:
    cooperativeMonitoring: majority_of_all
    interval: 2s
    module: mariadbmon
    name: mariadbmon-monitor
    params:
      auto_failover: "true"
      auto_rejoin: "true"
      switchover_on_low_disk_space: "true"
  podSecurityContext:
    fsGroup: 996
  replicas: 2
  requeueInterval: 10s
  servers:
  - address: mariadb-test-0.mariadb-test-internal.kube-system.svc.cluster.local
    name: mariadb-test-0
    port: 3306
    protocol: MariaDBBackend
  - address: mariadb-test-1.mariadb-test-internal.kube-system.svc.cluster.local
    name: mariadb-test-1
    port: 3306
    protocol: MariaDBBackend
  - address: mariadb-test-2.mariadb-test-internal.kube-system.svc.cluster.local
    name: mariadb-test-2
    port: 3306
    protocol: MariaDBBackend
  serviceAccountName: mariadb-test-maxscale
  services:
  - listener:
      name: rw-listener
      params:
        connection_metadata: tx_isolation=auto
      port: 3306
      protocol: MariaDBProtocol
    name: rw-router
    params:
      master_accept_reads: "true"
      max_replication_lag: 3s
      max_slave_connections: "255"
      transaction_replay: "true"
      transaction_replay_attempts: "10"
      transaction_replay_timeout: 5s
    router: readwritesplit
  - listener:
      name: rconn-master-router-listener
      port: 3307
      protocol: MariaDBProtocol
    name: rconn-master-router
    params:
      master_accept_reads: "true"
      max_replication_lag: 3s
      router_options: master
    router: readconnroute
  - listener:
      name: rconn-slave-router-listener
      port: 3308
      protocol: MariaDBProtocol
    name: rconn-slave-router
    params:
      max_replication_lag: 3s
      router_options: slave
    router: readconnroute

Expected behaviour
See #586 (comment).

Steps to reproduce the bug

  1. Create a MariaDB object with Maxscale enabled.
  2. Modify the switchover_on_low_disk_space parameter in the maxscale module configuration in MariaDB.
  3. Observe the behavior of the operator and the MaxScale object.

Environment details:

  • Kubernetes version: [v1.17.2]
  • mariadb-operator version: [0.0.27]
  • Install method: [helm]
  • Install flavor: [custom]

Additional context

mmontes11 commented

Hey there @luohaha3123 ! Thanks for reporting

The embedded MaxScale inside MariaDB is limited and does not currently support updates:
https://github.com/mariadb-operator/mariadb-operator/blob/main/docs/MAXSCALE.md#maxscale-embedded-in-mariadb

This is intended to quickly provision a non mutable MaxScale instance wiithout having to create an extra resource. I encourage you to create a dedicated MaxScale resource as described here:
https://github.com/mariadb-operator/mariadb-operator/blob/main/docs/MAXSCALE.md#maxscale-cr

In our side, we will make the spec.maxscale field of MariaDB inmutable to improve the feedback to the user.

Thanks!

luohaha3123 commented

@mmontes11 Thanks for your response. I have a question. If I have an external MaxScale for a replication-based MariaDB setup, will both of them handle failover and other functionalities simultaneously?

mmontes11 commented

This will be a good question for the MaxScale team. I encourage you to join our slack channel and ask in #help :
https://r.mariadb.com/join-community-slack

github-actions commented

This issue is stale because it has been open 30 days with no activity.

github-actions commented

This issue is stale because it has been open 60 days with no activity.

github-actions commented

This issue was closed because it has been stalled for 30 days with no activity.