PR #324 changed addresses for cluster nodes to use fqdn.
When upgrading from a cluster deployed for a revision previous to #324 , on rollback, querying for the cluster status, even though the instance is functional in the cluster, there are instanceerrors being reported, which will fail the pre-upgrade-check action step for online instance, since the count for online units take instanceErrors into account:
{
"mysql-k8s-0": {
"address": "mysql-k8s-0.mysql-k8s-endpoints:3306",
"instanceErrors": [
"ERROR: Metadata for this instance does not match hostname reported by instance (metadata=mysql-k8s-0.mysql-k8s-endpoints:3306, actual=mysql-k8s-0.mysql-k8s-endpoints.m6.svc.cluster.local:3306). Use rescan() to update the metadata.",
"ERROR: Metadata for this instance does not match X plugin port reported by instance (metadata=mysql-k8s-0.mysql-k8s-endpoints:33060, actual=mysql-k8s-0.mysql-k8s-endpoints.m6.svc.cluster.local:33060). Use rescan() to update the metadata."
],
"memberRole": "SECONDARY",
"mode": "R/O",
"readReplicas": {},
"replicationLagFromImmediateSource": "",
"replicationLagFromOriginalSource": "",
"role": "HA",
"status": "ONLINE",
"version": "8.0.34"
},
github-actions[bot]
commented
10 months ago
PR #324 changed addresses for cluster nodes to use fqdn.
When upgrading from a cluster deployed for a revision previous to #324 , on rollback, querying for the cluster status, even though the instance is functional in the cluster, there are
instanceerrorsbeing reported, which will fail thepre-upgrade-checkaction step for online instance, since the count for online units takeinstanceErrorsinto account: