Re: enq: WL - contention; plus kcrrrcp
- From: Ravi Madabhushanam <ravi.madabhushanam@xxxxxxxxx>
- To: gus.spier@xxxxxxxxx
- Date: Sat, 24 Jul 2010 15:43:19 +0530
We also had very bad experiences with enq-WL. In our case, this often occurs
on the archiver background process while the RMAN backup was running.
Particular statement that was causing all these issues were, "alter system
archivelog current". In particular the issue was specific to archivet
background process not able to complete the archiving. Does any these sound
similar to you?
After following up with Oracle for good amount of time, its finally observed
that there is a bug in 10g ASM due to which archiver will go into hung state
resulting in bad performance of all instances/databases across the cluster.
Our environment was 10gR2 DB, 10gR2 ASM. on SLES 10.
~Ravi.M
On Fri, Jul 23, 2010 at 11:11 PM, Gus Spier <gus.spier@xxxxxxxxx> wrote:
> It simply has to be a Friday ...
>
> Solaris 10
> Oracle 10.2.0.2 EE
>
> A member of the end-user community calls in a trouble ticket, "The database
> is going awfully slow!" No further information or specifics are available
> ... just "the database is going awfully slow!"
>
> addmrpt alerts us to an issue "enq: WL - contention waits in service
> SYS$BACKGROUND" and follows that up with an issue "kcrrrcp waits in
> SYS$BACKGROUND"
>
> A quick query to V$EVENT_NAMES tells me that enq: WL refers to log writing
> (do I assume this archiving the REDO logs?) ... kcrrrcp is simply listed as
> "Other".
>
> Google has been able to offer very little: a brief reference in our
> archives here that deal with an RMAN bug.
>
> Has anyone seen these types of things before? Is there some other
> documentation that I should be looking in? Wasn't there once upopn a time a
> Working Aid that decrypted all the kg*, kc*, kl* modules with at least a
> good guess about their function and usage.
>
> Any assistance will appreciated.
>
> regards,
> Gus
>
Other related posts:
