Location of ERRORLOG and SQLAGENNT.OUT

SQL Server 2000, Windows 2003 R2, Cluster 5.2
We're setting up a new cluster and I would like to move the LOGFILE and
SQLAGENT.OUT file off of the drive(s) that contain .ldf & .mdf files. My
reason is that these files will contribute to drive fragmentation (also, I
would like it to be a simpler path to get to the files, something right off
of the drive root).
Now, I've never done this on a cluster before and our Windows admin is
telling me that these logs are somehow used by the cluster administrator to
do some part of the failover. I'm unclear how.
Anyway, The drives I have available are:
C: OS (not shared)
E: Apps (SQL Server Install/not shared)
Q: Quorum
L: Logs (shared)
S: Data (shared)
R: Backup (shared)
The ERRORLOG & SQLAGENT.OUT files are currently on S: under the "S:\Program
Files\Microsoft SQL Server\MSSQL\LOG" path (same disk as the data files) and
I would like to place them on the E: drive (which is not shared).
I can see that this would cause me to not have log entries from the other
server immediatly available to me on a changeover as it wouldn't be on a
shared disk -- and I don't know if that is a bad thing, or not.
Can I, or should I, move these log files off the disk with the data files?
Thanks,
Jay
These log files hardly cause any significant fragmentation that would have a
significant impact on the database performance.
I have never tried to move the log files to local non-shared disks, and will
probably never try it since that is not the way it is supposed to work.
Even if you managed to relocate the log files to E, consider this. When your
instance is running on node A, and the logs would be on the E drive on node
A. If then node A is hosed, your SQL instance will fail over to another node,
but your log files. So you won't have access to the log files for
troubleshooting. Not good! Okay, the cluster service may manage to replicate
some of the errorlog entries that are also writtne to the app eventlog to the
other nodes, but the completeness of that is not guaranteed when there is a
sudden failure.
Don't do it!
Linchi
"Jay" wrote:

> SQL Server 2000, Windows 2003 R2, Cluster 5.2
> We're setting up a new cluster and I would like to move the LOGFILE and
> SQLAGENT.OUT file off of the drive(s) that contain .ldf & .mdf files. My
> reason is that these files will contribute to drive fragmentation (also, I
> would like it to be a simpler path to get to the files, something right off
> of the drive root).
> Now, I've never done this on a cluster before and our Windows admin is
> telling me that these logs are somehow used by the cluster administrator to
> do some part of the failover. I'm unclear how.
> Anyway, The drives I have available are:
> C: OS (not shared)
> E: Apps (SQL Server Install/not shared)
> Q: Quorum
> L: Logs (shared)
> S: Data (shared)
> R: Backup (shared)
> The ERRORLOG & SQLAGENT.OUT files are currently on S: under the "S:\Program
> Files\Microsoft SQL Server\MSSQL\LOG" path (same disk as the data files) and
> I would like to place them on the E: drive (which is not shared).
> I can see that this would cause me to not have log entries from the other
> server immediatly available to me on a changeover as it wouldn't be on a
> shared disk -- and I don't know if that is a bad thing, or not.
> Can I, or should I, move these log files off the disk with the data files?
> Thanks,
> Jay
>
>
|||Point taken, thanks.
"Linchi Shea" <LinchiShea@.discussions.microsoft.com> wrote in message
news:2B7D6FFA-DAF5-4F1A-8479-95AD50156DE7@.microsoft.com...[vbcol=seagreen]
> These log files hardly cause any significant fragmentation that would have
> a
> significant impact on the database performance.
> I have never tried to move the log files to local non-shared disks, and
> will
> probably never try it since that is not the way it is supposed to work.
> Even if you managed to relocate the log files to E, consider this. When
> your
> instance is running on node A, and the logs would be on the E drive on
> node
> A. If then node A is hosed, your SQL instance will fail over to another
> node,
> but your log files. So you won't have access to the log files for
> troubleshooting. Not good! Okay, the cluster service may manage to
> replicate
> some of the errorlog entries that are also writtne to the app eventlog to
> the
> other nodes, but the completeness of that is not guaranteed when there is
> a
> sudden failure.
> Don't do it!
> Linchi
> "Jay" wrote:
sql