Last updated on SEPTEMBER 16, 2016
Applies to:Oracle Database - Enterprise Edition - Version 18.104.22.168 to 22.214.171.124 [Release 11.2]
Information in this document applies to any platform.
The database performance is very slow, and there are spikes of slowness. The database performance eventually may recover after some time (often lasts more than an hour).
1) The AWR report shows high gc waits including "gc cr/current grant .. " waits.
The "gc cr/current grant .. " waits usually indicate busy lms processes as the the grant function involves transmission of very small packets, so the network is not likely the cause.
Top 5 Timed Foreground Events
Event Waits Time(s) Avg wait (ms) % DB time Wait Class
gc cr grant 2-way 759,706 54,668 72 18.46 Cluster
DB CPU 40,218 13.58
gc current grant busy 75,783 27,518 363 9.29 Cluster
gc current grant 2-way 237,492 24,475 103 8.26 Cluster
gc current block busy 51,179 21,462 419 7.25 Cluster
2) The top and ps outputs of OSWatcher show that lms is taking 100% of one CPU.
3) A trace file (usually lmon) shows one or more lms processes are not moved for many seconds although the problem lms is not waiting for any resource. The trace file shows the the short stack of the lms includes ktrgcm.
LMS0 (ospid: 29316) has not moved for 76 sec (1384485973.1384485897)
kjfmGCR_HBCheckAll: LMS0 (ospid: 29316) has status 2
: Not in wait; last wait ended 0 secs ago.
: last wait_id 238514 at 'gcs remote message'.
4) in AWR report, the stats "data blocks consistent reads - undo records applied" is much larger (in millions) compare with other stats
Sign In with your My Oracle Support account
Don't have a My Oracle Support account? Click to get started
Million Knowledge Articles and hundreds of Community platforms