Re: Calling in the Cavalry - Hanging 8i database

>>Does the replication 'hang' as well? Is the replicated database >>affected in any way? Is your network causing problems with regard to

>>replication?

All active sessions are completely frozen including the replication. The replication takes place over a link between UK and USA. It has been setup so that transactions are pushed to each side remotely every minute. We have experienced periods in the past where the network has been down for a period of time without it causing a dbase hang. When this occurs transactions at each side are simply queued until the link is restored and they can be processed (there are also entries in the alert log). It is feasible that the link may have spikes in it during the day but we have other applications (primarily web-enabled) running over it and there are rarely problems. It is worth investigating though.

> The system then recovers itself. However, the hanging does not leave
> any trace files or entries in the alert log.

>>If it isn't a database issue, per se, you won't have any alert log >>messages.

Yes I agree. This is why I had the Solaris software / hardware checked out.

> I have had the hardware checked out by a Solaris consultant and he
> could find nothing wrong. There are also some other apps running on
the
> unix box which have never been affected.

>>It doesn't necessarily indicate a hardware problem; you are replicating
>>this to a remote site, which requires a working network connection. As
>>I asked previously is the replication affected when this 'hang' occurs?
>> Is the network still properly functioning during these 'outages'?

I will need to wait until there is a another hang and then test replication from the USA side by pushing out a transaction. This is not possible from UK (hanging) side as all connections are frozen.

Thanks for the reply - most appreciated.

• Hide quoted text -
• Show quoted text -

> To try and find the cause of the problem, I have worked closely with
> Oracle support via many tars and provided numerous systemstate level
10
> dumps which have highlighted nothing. I've also had an Oracle
> consultant onsite who failed to uncover any irregularities.

> The box is not particularly overworked and the hanging is not linked
to
> periods of heavy activity. Many of the database connections are from
a
> Weblogic box which runs our internal java application.

> I'm beginning to think that I need to completely rebuild the box.
> However, I would like one last attempt to determine what is causing
the
> hanging.

Do not stop at the hardware and the database. Other areas may affect this behaviour.

• Hide quoted text -
• Show quoted text -

> The instance runs in archivelogmode and all redo logs are 25MB in
size.
> There are no issues with disk space etc. as this is all closely
> monitored.

> Could anybody suggest any appropriate course of action that I could
> follow to start the investigation. I'm largely a developer/dba and
not
> hugely experienced.

> Please advise of any info required concerning the instance
> configuration and I will post asap.

> Thanks in advance to anybody who can point me in the right direction.

> regards
> John
Received on Sat May 14 2005 - 16:13:00 CDT