Terribly wrong Crash

[Jai Gupta]

PCX is crashing without giving backtrace. We have tried exporting all data using mysqldump and doing a clean import.

Percona-XtraDB-Cluster-client-56-5.6.20-25.7.888.el6.x86_64
Percona-XtraDB-Cluster-galera-3-debuginfo-3.7-1.3254.rhel6.x86_64
Percona-XtraDB-Cluster-shared-56-5.6.20-25.7.888.el6.x86_64
Percona-XtraDB-Cluster-server-56-5.6.20-25.7.888.el6.x86_64
Percona-XtraDB-Cluster-test-56-5.6.20-25.7.888.el6.x86_64
Percona-XtraDB-Cluster-full-56-5.6.20-25.7.888.el6.x86_64
Percona-Server-shared-compat-5.5.35-rel33.0.611.rhel6.x86_64
Percona-XtraDB-Cluster-56-debuginfo-5.6.20-25.7.888.el6.x86_64
Percona-XtraDB-Cluster-galera-3-3.7-1.3254.rhel6.x86_64
Percona-XtraDB-Cluster-garbd-3-3.7-1.3254.rhel6.x86_64

Because there was no backtrace we enabled general query log to troubleshoot.

[Jai Gupta (jai-g)]

Another node crashed.

I am attaching Query Log, Mysql Log and my.cnf from this node.

It seems that crash occurs when user session data is stored in database, we are using open source PHP application Moodle.

[Jai Gupta (jai-g)]

Could Bug #1367562 be related?

[Raghavendra D Prabhu (raghavendra-prabhu)]

@Jai,

Are ou still see this issue? I also see you are using RELEASE_LOCK which is not supported.

Also, crash trace is missing from the logs.

[Jai Gupta (jai-g)]

We now don't round robin connections between multiple DB nodes. In place, we use single Node for all connections and only use other Nodes as failover. This is working for us for now.

If we start using round robin, node starts crashing.

There is no crash trace, it just crashes real bad.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PXC-1738