I am new to Percona XtraDB Cluster and exploring. Couple of days before faced an issue with the cluster which brought down entire cluster. I did go though the logs and suspect the issue was caused by connection issue of nodes to peers. The cluster contains 3 nodes and application talks with the database via Haproxy. Need some expert advice if it was a connection issue or something else that I am missing. I rebuilt the cluster as this was our QA server by bootstrapping one node. Need to know if this is caused by a configuration issue or not so that similar scenario does not happen on production servers.
I am attaching the logs from all the nodes.
Node_Logs.txt (34 KB)
From the logs (attached in previous post), I think the issue was due to certificate failure during the processing of DLL statement DROP TABLE IF EXISTS under TOI method. The table being dropped is a temporary table and I have already put the clause IF EXISTS with drop. So, even if the table does not exists in any other node it should not throw an error. In normal MySQL it just throws a warning message.
Is this a bug in Percona XtraDB cluster ? Does XtraDB cluster replicate temporary tables which only have a scope at session level. Was the error because of me not using explicitly drop temporary table ?
Please let me know if my observations are correct or was the issue due to something else ?
PXC does not replicate temporary tables. There have been some changes to this code in this area, so the PXC version will make a difference. What version of PXC are you using?
The version of PXC that I use is
Server version: 5.6.35-80.0-56-log Percona XtraDB Cluster (GPL), Release rel80.0, Revision 62752df, WSREP version 26.20, wsrep_26.20