Dear all:
Things are improving, but there are still some unresolved issues.
Status so far is that the backup problem seems to be solved by
reorganising NSR clients. This system used to run ~25 ASE
services that was backed up by a server outside the cluster. Now
we have 'consolidated' the clients into 'local' backups, using the
current NSR server node as client to avoid network transfer. This
have fixed the backup, but another story is the NSR installation
itself...
The holiday season has started and that may explain why users
now are happy about performance, because we still see high
cpu load on the systems that we did not see before. During this week we
will have a closer look at this, and pay special attention to network
setup, including switches and adapter settings.
The responses I got touches NSR and network settings, and I also
got hints about patch 399 in PK3 causing heavy load. Since this
is a 'jumbo' patch and since things seems to settle a bit, I prefer
to wait on this. Thanks to:
Joe Fletcher, Kevin Jones, Manish Vashi, Raul Sossa S.
Alex Gorbachev, Kevin Criss
If I get more info, I will update my summary.
Happy holidays!
Regards,
Hallstein
Original posting:
> Fellows,
>
> The title may not be true, but on a 3 node TruCluster we have
> seen a serious slowdown after the latest upgrade which
> included the following:
>
> - Upgrade cache on two dual HSG80's, from 64 to 256 MB,
> both pairs in multibus failover.
> - Doubled memory on one node (ES40) to 4 GB
> - Upgrade to latest FW on KGPSA's (dual on all nodes)
> - Installed Patch-kit 3 for Tru64 / TruCluster V5.1
>
> Almost immediately after the upgrade the slowdown was seen, this
> cluster runs a bunch of Oracle DB's and Oracle App's. Soon afterwards
> a slowdown was also seen on backup which is a clustered Networker 6.01
> installation on two of the nodes. Here it is possible to get figures;
> before the upgrade, the backup was running at about 6 MB/s, but now
> we are seeing close to 1 MB/s. This lead to problems with the backup
> window, and at the moment backup is the biggest problem. Compaq has
> the ball, and so far they have reduced the HSG80's cache back
> to the previous
> amount, and deinstalled the ES40 memory upgrade, but the
> slowdown persist.
>
> The next thing on the list may be rolling back to PK1, or may be going
> back on KGPSA FW, but all what I have read so far indicate
> that all we
> did in the first place should improve performance.
>
> I feel a bit stuck here...
>
> Regards
>
> Hallstein Lohre
> Alpha System AS
> Trondheim
> Norway
>
Received on Mon Jul 02 2001 - 12:23:38 NZST