Predefined labels v2.1.0

These predefined labels are managed by the EDB Postgres Distributed for Kubernetes operator.

• k8s.pgd.enterprisedb.io/certificateType

  Indicates the type of the certificates. replication indicates a certificate to be used to authenticate the replication client. server indicates a certificate to be used for server authentication.

• k8s.pgd.enterprisedb.io/group

  Name of the PGDGroup that the resource belongs to. Added to cluster or instance resources.

• k8s.pgd.enterprisedb.io/isWitnessService

  Indicates a service is for a witness node.

• k8s.pgd.enterprisedb.io/nodeRoutable

  Indicates that the readiness probe of this node monitors the Connection Manager. Nodes that don't carry the label, or that carry it with the value false, aren't included in the group's proxy and proxy-r services, and their readiness probe monitors only the PostgreSQL server. The label is added when a new node joins the cluster, and on existing nodes after upgrading the operator from a version before 2.1.0. A parting node carries false. The disableConnMgrReadinessProbe annotation on the PGDGroup disables readiness probe management for the Connection Manager and for the proxy services of the group. The operator writes this label; it's not meant to be set by hand.

• k8s.pgd.enterprisedb.io/parentGroupUUID

  Added to node and instance pods to show the UUID of the global PGD group, used to group the PGD clusters in different regions.

• k8s.pgd.enterprisedb.io/type

  Type of the resource added to cluster or instance resources, usually node.

• k8s.pgd.enterprisedb.io/workloadType

  Indicates the workload type of the resource added to cluster or instance resources. pgd-node-data indicates data node; pgd-node-witness a witness node; proxy-svc for PGD Proxy service; group-svc for PGD group service to communicate with any node in the PGDGroup; node-svc is a service created from the CNP service template; scheduled-backup is added to scheduledBackup resources; bootstrap-cross-location-pgd-group is added to the pod that creates a cross-location PGD group; pgd-node-restore is added to the pod that starts the node restore process.

Predefined annotations

• k8s.pgd.enterprisedb.io/dirtyMetadata

  Set in a CNP cluster that has been generated from a backup and needs to have its metadata cleaned up before creating the PGD node. This is written by the restore job.

• k8s.pgd.enterprisedb.io/disableConnMgrReadinessProbe

  Set to true in the PGDGroup to prevent the readiness probe of its data nodes from monitoring the Connection Manager, reverting to the operator behavior of versions before 2.1.0. The data node pods are recreated once, node by node, when this annotation is set or removed.

• k8s.pgd.enterprisedb.io/hash

  Holds the hash of a certain part of PGDGroup spec that's utilized in various entities like Cluster, ScheduledBackup, StatefulSet, and Service (node, group and proxy service) to determine if any updates are required for the corresponding resources.

• k8s.pgd.enterprisedb.io/unsafeIgnoreRaftCheck

  Set to true on a PGDGroup to let the operator restart nodes while the raft has no spare member. It permits a restart with no quorum margin, so it's a last resort: remove it as soon as the group is healthy. See Restart held, raft has no spare member for the recovery options to try first.

• k8s.pgd.enterprisedb.io/latestCleanupExecuted

  Set in the PGDGroup to indicate that the cleanup was executed.

• k8s.pgd.enterprisedb.io/node

  Contains the name of the node for which a certain certificate was generated. Added to the certificate resources.

• k8s.pgd.enterprisedb.io/nodeRestartHash

  Stores the hash of the CNP configuration in PGDGroup. A restart is needed when the configuration is changed.

• k8s.pgd.enterprisedb.io/noFinalizers

  Set in the PGDGroup with value true to skip the finalizer execution. For internal use only.

• k8s.pgd.enterprisedb.io/pause

  Set in the PGDGroup to pause a PGDGroup.

• k8s.pgd.enterprisedb.io/part

  Set in the PGDGroup with value on to part all the nodes in the PGDGroup. Once all nodes are parted, each CNP Cluster is annotated with value parted. This annotation is irreversible and blocks spec changes on the PGDGroup.

• k8s.pgd.enterprisedb.io/recoverabilityPointsByMethod

  Set in the PGDGroup to store the CNP cluster's first recoverability points by method in a tamper-proof place.

• k8s.pgd.enterprisedb.io/reconciliationLoop

  Set in the PGDGroup with disabled to temporarily disable the reconciliation from the operator.

• k8s.pgd.enterprisedb.io/remoteTarget

  Set in the PGDGroup during remote join by the bootstrap job, to indicate the remote node to physically join.

• k8s.pgd.enterprisedb.io/seedingServer

  Set in the PGDGroup to indicate to the operator which server to restore. This is written by the restore job.

• k8s.pgd.enterprisedb.io/seedingSnapshots

  Set in the PGDGroup to indicate to the operator which snapshots to restore. This is written by the restore job.

• k8s.pgd.enterprisedb.io/useBarmanCloudPlugin

  Set in the PGDGroup with value true to instruct the operator to enable the barman-cloud plugin. Once the plugin is enabled, a new sidecar container will be created for the instance pod. The sidecar container will manage the barman base backup and WAL archiving, rather than the PostgreSQL container. Keep in mind that the barman-cloud plugin must be installed in the PG4K operator namespace beforehand, and a restart of instance pod is required for these changes to take effect.