Oracle 10g:
I see very very long checkpoints in alert logs:
Mon Mar 23 12:35:57 2009
Beginning log switch checkpoint up to RBA [0x2001.2.10], SCN: 2667813010
Mon Mar 23 12:35:57 2009
Thread 1 advanced to log sequence 8193 (LGWR switch)
Current log# 3 seq# 8193 mem# 0: /u05/oradata/perf/redo03a.log
Current log# 3 seq# 8193 mem# 1: /u06/oradata/perf/redo03b.log
Mon Mar 23 12:41:02 2009
Completed checkpoint up to RBA [0x2001.2.10], SCN: 2667813010
Our redo logs are 2gb and have 6 of those.
I am not sure why checkpoints are so long. Also, does this affect the requests trying to read-write to DB?
I am seeing high response time on our customer transactions when backup runs and one of the things I see is high checkpoints. But I am not sure if that would cause 20x high response time even though we do it at night time when the traffic is relatively low. There is something wrong and I am not sure if it's chekpoints, redo logs or something else.
Edited by: user628400 on Mar 23, 2009 5:50 PM
Edited by: user628400 on Mar 24, 2009 10:09 AM