[prev in list] [next in list] [prev in thread] [next in thread] 

List:       veritas-ha
Subject:    RE: [Veritas-ha] mount agent impacting system load
From:       "Van Atta, Tom" <Tom.VanAtta () scotts ! com>
Date:       2002-05-31 21:03:24
[Download RAW message or body]

In my case my problem was with the way the Oracle resource was set up. It
was causing VCS to try to mount all 70 filesystems at the same time. This
did put an adverse load on the system and affected LLT.

I ended up creating dependencies between the mounts, so only 5 or 10
filesystem mount commands are issued by VCS at any given time. This cleared
up my issue and had no impact on the startup time of the bringing the
database online.

If you are seeing this with no service groups defined then I would suspect
you have some other issue at play. I would suggest opening a support case
with Veritas - I'm probably not a good resource to try to resolve or
speculate on what might be causing your problem. I will say I've gotten
great support from other (more knowledgeable) members of this mail list.

Hope this helps,

-Tom

-----Original Message-----
From: paul [mailto:paul@hessels.ca]
Sent: Friday, May 31, 2002 3:14 PM
To: Van Atta, Tom
Cc: veritas-ha@mailman.eng.auburn.edu
Subject: Re: [Veritas-ha] mount agent impacting system load


Tom, I am seeing these same messages, and my load isn't very high at all.

I have no clue to the problem.  I get this errer regardless of the number 
of service groups... even zero service groups.

Anyone have any clues?

On Mon, 4 Feb 2002, Van Atta, Tom wrote:

> I am running VCS 1.3p3 with the oracle agent. The oracle database contains
a
> large number of data files spread across approx. 70 filesystems. I've
> noticed when bringing the Oracle service group online or offline, the
server
> becomes unresponsive during the online/offline of the mount agent. When
this
> occurs, several monitor processes report they have timed out, and these
> messages regarding LLT appear in the system log:
> 
> 
> 	Feb  3 16:15:03 sapprd llt: [ID 687274 kern.notice] LLT:10035: timer
> not called for 120 ticks
> 	Feb  3 16:16:05 sapprd llt: [ID 687274 kern.notice] LLT:10035: timer
> not called for 129 ticks
> 	Feb  3 16:16:38 sapprd llt: [ID 687274 kern.notice] LLT:10035: timer
> not called for 127 ticks
> 	Feb  3 16:16:42 sapprd llt: [ID 687274 kern.notice] LLT:10035: timer
> not called for 109 ticks
> 
> 
> In looking over cpu utilization during this transition, system time jumps
to
> 50 - 70% during this transition.
> 
> Due to the large number of mount points, I had increased the number of
> threads from 10 to 20 for the mount agent. After observing the above
impact
> on the system, I decided to drop the number of threads down to 12.
> 
> Am  I correct in assuming that the number of threads caused the excessive
> load, or should I look elsewhere? As a side note, I've been running a test
> cluster with the same configuration/threads and did not notice the same
> problem.
> 
> Thanks,
> 
> -Tom
> _______________________________________________
> Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha
> 

-- 
Just my 2 cents... if I keep spending money like this I'm soon going to be
poor!




_______________________________________________
Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha
[prev in list] [next in list] [prev in thread] [next in thread]