Details
-
Bug
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
10.5.12
-
None
-
None
-
None
-
centos7.5
Mariadb 10.5.12
Galera provider version: 26.4.9
Description
Happy new year!
We have a gelera cluster with 3 nodes in production, and we upgraded to 10.5.12 recently from 10.3.10.
We provide a keepAlived +Haproxy to access the DB clutster.
The galera cluster would hang (can only select and unable to do update or delete) after one node reboots occationally.
The application gets the error "Lock wait timeout exceeded; try restarting transaction" and there is no error log in mysql logs. I can do some select in termincal but any update and delete sql would hang.
And *wsrep_local_state_comment *on 3 nodes are "Synced". *Wsrep_last_commited *on 3 nodes are static and would not go forward any more. And one is lower than other two.
To recover the cluster, it would work to restart the mariadb instance which has differrent wsrep_last_commited or reboot the cluster with --wsrep-new-cluster.
The possibility would be much higher when one node is poweroff, and reboot another node.
Any advice would appreciate, thanks!