Shunki

While reading restore.go, I noticed that the option -skipBinlog=true was included. Thus, binlog was disabled during the restore. https://github.com/cybozu-go/moco/blob/539a90d2e329589f4d05612080584a30820b32b3/pkg/bkop/restore.go#L36

@ymmt2005 How about delaying the execution of `CHANGE MASTER TO`? Compare `gtid_purged` of the new primary with `gtid_executed` of the replica using `gtid_subtract()` and wait until there is no difference....

@ymmt2005 The problem is that the primary pod recreated by CLONE has no binlogs prior to the time of CLONE. If all pods are re-created with CLONE, the process is...

@ymmt2005 Ahhhhh,.. the procedure in the text was a pseudo-reproduction of the replication delay. Sorry for the confusion. What actually happened was that the `nodeSelector` change caused all pods (and...

The problem is that the switching is not completed within the PreStopSeconds time. Solution will be considered separately.

@rassie These logs seem to occur during the construction of MySQLCluster. Do these logs stop appearing once the construction of MySQLCluster `myapp` is completed? Or does the construction fail to...

@mhkarimi1383 I tried to reproduce this locally. I do see the same log: ``` { "level": "error", "ts": "2025-09-16T05:48:33.979Z", "msg": "failed to update status", "controller": "mysqlcluster", "controllerGroup": "moco.cybozu.com", "controllerKind": "MySQLCluster",...

@mhkarimi1383 I was debugging why increasing MySQLCluster.spec.volumeClaimTemplates[].resources.requests.storage (e.g., 5Gi → 10Gi) didn’t raise the PVC spec.resources.requests.storage. It turned out to be caused by putting an empty resources map on the...

@mhkarimi1383 This is expected behavior. MOCO prioritizes StatefulSet reconciliation (resolving Pod template diffs / rolling updates). As a result, PVC reconciliation only runs once all Pods are Healthy/Ready. If nodes/Pods...

@YZ775 Also add `MySQL 8.4.2` https://dev.mysql.com/doc/relnotes/mysql/8.4/en/news-8-4-2.html