Last updated on SEPTEMBER 14, 2016
Applies to:Oracle Communications Messaging Server - Version 6.3 and later
Information in this document applies to any platform.
Performance of syscalls like write() and fdsync() on file systems residing on ZFS are slow, but asvc_t in iostat on the underlying devices looks fine.
Best Practice recommendations for Messaging Server is that the "average service time of active transactions, in milliseconds" (the asvc_t column of iostat -x output) needs to be less than 20 milliseconds. Above that is the point where disk performance begins to have a cascade effect on performance.
So when any part of Messaging Server performance begins to degrade, one of the first steps has always been to look at iostat to see if any of the disks are over loaded.
But sometimes scripts like delivery-perf.d, delivery-perf-detail.d, and imapd-txns-cmds-IO.d -- see Monitor Filesystem IO Calls for Messaging Server (Doc ID 1351483.1) -- report write() or fdsync()syscalls taking unreasonably long times but iostat shows the disks are fine.
Load may have increased gradually over time and finally reached a breaking point.
Recent patch/upgrade from Solaris 10 update 8 or earlier to Solaris 10 update 9 or later.
Sign In with your My Oracle Support account
Don't have a My Oracle Support account? Click to get started
My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms