RE: Error Log file location causing Failover failure?
- From: "GNocent" <GNocent@xxxxxxxxxxxxxxxxxxxxxxxxx>
- Date: Tue, 14 Feb 2006 05:40:30 -0800
Hi,
For the SQLServer part, the "filesystem full" alert is not a fatal error, so
the instance won't fail and the cluster wont move. You can have a database
growth that is blocked by the lack of disk space, and failing over wont
change that ...
What I undertsand, is that the tool you installed is not SQLServer based,
but node based. The difference is important.
* If it is purely SQLServer based, it has to work on the virtual instance,
on the virtual name, with the virtual IP, and on the shared disks included in
the instance's group.
* If it is node based, you will make 1 setup on each node, each one will be
independant, and all files will have to be installed on local disks (not
shared ones).
If you mix these two things (node based installation but with files on
shared disks), your software won't be able to run correctly, and, more
dangerous, can affect the stability of you cluster.
Managing a SQLServer requires minimal knowledge of the product, and more
knowledge if its clustered as some operations need some more precautions. If
you are the only person in charge of this system, I advise you to ask you
hierarchy for a training on SQLServer and on MS Clustering (or at least some
good books), as that's only when things goes wrong that you realize that some
more knowledge would have save you from the situation, and not the opposite
....
Good luck !
Guillaume.
=======================================
"Nancy Lytle" wrote:
I am kind of the defacto SQL DBA in our small company. We have installed.
Idera's SQL Diagnostic Manager to assist me in monitoring the SQL Servers.
Yesterday I received the following email from our Net Admin.
Evidently we had one of our clustered SQL Servers fail because it ran out of
disk space, the cluster then did not fail over.
The reporting software you have installed has a hook in place for when sqlmessage
starts that causes the sql server to look for a particular log file in order
to start up. This logfile is on the shared disk resource which gets moved
from one node to the next. The problem with this is it looks for the shared
disk using \\SQLDat01\ which only works if the disks are on that node. It
needs to be changed to \\SQLDataclst\. This prevented the cluster from
failing over last night.
end of message
Now, I don't think Idera's DM installed any hook. But I know very little
about clusters and SQL Server. Is there a "special" location where the
error logs need to be placed, and if so, how do you go about changing the
location of the error logs?
Hope this makes sense.
Any and all help appreciated.
Nancy
- References:
- Error Log file location causing Failover failure?
- From: Nancy Lytle
- Error Log file location causing Failover failure?
- Prev by Date: Re: Error Log file location causing Failover failure?
- Next by Date: Re: SQL 2005 Cluster with 2K
- Previous by thread: Re: Error Log file location causing Failover failure?
- Next by thread: Adding new nodes to existing SQL 2000 cluster - Problems problems and more problems
- Index(es):
Relevant Pages
|
