enkhbold Messages: 36 Registered: July 2005 Location: Mongolia
Member
Our database having a problem relating to LGWR.
Monitoring (Spotlight on windows) displays "average time to sync a Redo Log Entry alarm activated when the time taken to write redo log entries during a commit exceeds a threshold".
We thinking it is relating to Redo Log buffer size and writing process.
Redo Lug buffer size is 30MB and Redo files are 200MB each.
Database is Oracle10g RAC 64 bit. And, we allocated dedicated disk to REDO LOG files at the EMC storage.
please advice me on redo sizing and this alarm how to fix?
please share with me experience.
halfydba Messages: 20 Registered: November 2007 Location: Australia
Junior Member
1. "Redo Lug buffer size is 30MB"
Redo log buffer or log_buffer size of 30MB is excessive. This should probably no larger than 1MB as there is no real benefit of having a larger value.
-> Default Maximum is 512 KB or 128 KB * CPU_COUNT, whichever is greater.
Or else you could be causing yourself issue when writing the buffer to your online redo e.g. log file sync issues.
2. "And, we allocated dedicated disk to REDO LOG files at the EMC storage." are you using raid 1? raid 5 is no good for online redo logs when there is reasonable system activity.
Changing these will will probably help your log file sync issues
]
Members
Search
Help
Register
Login
Home
