Poor Performance On Just One Node

January 12, 2014

Just the LOAD node in the DPF cluster, which is used by the batch jobs to perform LOAD’s was slow compared to the other servers

  • Connects to the database would take over 50secs (Other servers under a second)
  • Queries were running for over 50Secs while the same query will be back in couple of seconds on other servers
  • db2 get dbm cfg would show the same symptom

Just about any db2 commands, even db2pd would show this symptom. While OS related commands work ok

Server resources(CPU,Mmory,SWAP I/O and Network) were not being taxed, nothing in the db2diag.log, notify log or /var/log/messages would give any clue on the situation either.

After having no luck in the DB2 area, network team was involved and their trace showed a traffic to the LDAP server right before i see the commands return. Whatever command i execute is waiting on something for the first 50secs before contacting the LDAP server.

This helped a lot and I started looking at LDAP related files used by DB2 and noticed the /etc/nsswitch.conf file used for name resolution is empty, copying it over from another working server fixed it.


tags: , , ,
posted in DB2 by Visu

Follow comments via the RSS Feed | Leave a comment | Trackback URL

Leave Your Comment

Powered by Wordpress and MySQL. Theme by Shlomi Noach, openark.org