LGWR consumes 100% CPU
Last updated on JULY 15, 2018
Applies to:Oracle Database - Enterprise Edition - Version 10.1.0.2 to 10.2.0.5 [Release 10.1 to 10.2]
Information in this document applies to any platform.
Log Writer Consumes 100% CPU with the instance going for Hang and the system goes for a freeze .
You might see the following symptoms :
--> The existing logged in user cannot do anything. New users will not be able to get connected.
--> The CPU on the DB Tier where the LGWR process exist shoots to 100%
--> When the CPU comes down, and no blocks are there with respect to LGWR, the problem goes away.
--> When the LGWR blocks the user sessions, system is in a freeze state and no new logins are allowed.
The following messages apper in the alert log .
WARNING: inbound connection timed out (ORA-3136)
--> Querying v$sessoions shows that LGWR is blocking some of the sessions
SQL> SELECT sid, blocking_session, username, blocking_session_status FROM v$session ORDER BY blocking_session ;
<Bug:2656965> LGWR SPINS AND CONSUMES ALL CPU AND INSTANCE HANGS
The workaround '_lgwr_async_io=false' to prevent lgwr from spinning did not fixed the issue . The problem may manifest even after _lgwr_async_io = false has been set and instance bounced .
You can view the file systems (and the options) mounted with the mount command :
#device device mount FS fsck mount mount
#to mount to fsck point type pass at boot options
/dev/dsk/c2t2d0s0 /dev/rdsk/c2t2d0s0 /d16 ufs 2 yes -
/dev/dsk/c2t3d0s0 /dev/rdsk/c2t3d0s0 /d17 ufs 2 yes -
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