0

I browsed a lot about above exception, and got some doubt for one of the servers I' am looking into. I have Websphere Application Server 8 running with MySQL 5.5, all works fine. Until one day when the connections start failing on runtime and DB is running fine and no networking issues. Websphere needs a restart to recover from the situation. WebSphere logs "communication failure messages, caused by EOF: Can not read response from Server" thourgh out the period.

Could it be due to having default (1800 seconds) time set for Unused connections in websphere connection pool and low default (30s and 60s) net_read_timeout and net_write_timeout for Mysql?

MySql Timeouts: connect_timeout :10s interactive timeout : 28800s wait_timeout : 28800s net_read_timeout : 30s net_write_timeout : 60s

WebSphere timeouts: Unused Timeout :1800s Reap time : 180s minimum connections :10 max connections : 250 connection timeout : 180s

Thanks!

simer
  • 23
  • 4

1 Answers1

0

Yes, that is a possibility, the easiest way to know if it is the issue is to increase the net_[read|write]_timeout settings to 1815 seconds on mySQL side and see if it reoccurs. Alternatively, you could lower the unused t/o value on WebSphere to 25 sec. This would cause an inactive connection to be reaped before mySQL times it out. If you post the exceptions from your log, it might offer more insight.

F Rowe
  • 2,042
  • 1
  • 11
  • 12
  • Thank you for your response, you can find the stack trace of exception on this online notepad. Additional information it happens when the DB side is brought up before the application server. [https://shrib.com/see/NWGPOm7JhaOmN80nB9rdnJsskS5GNSRaoXq3AnChERKSQB5RAn?v=pr&s=m] – simer Mar 11 '16 at 16:25