Don't know whether this will help or not, but the general rule-of-thumb is
that the cache size should be about 20 percent larger than the number of
objects.
ovwdb keeps the entire object database in memory and what you want to avoid
is him having to realloc bigger and bigger chunks of memory, and then
copying what he has into the new chunk before freeing the old one. It
works, but it usually takes a much longer time than having him get the
right amount to begin with.
James Shanks
Level 3 Support for Tivoli NetView for UNIX and Windows
Tivoli Software / IBM Software Group
"Bursik, Scott
{PBSG}"
<Scott.Bursik@pbs To
g.com> <nv-l@lists.us.ibm.com>
Sent by: cc
owner-nv-l@lists.
us.ibm.com Subject
RE: [nv-l] nvUtil and nvcold
06/01/2006 12:26
PM
Please respond to
nv-l@lists.us.ibm
.com
Here is what I have found:
ovobjprint | head -1
Number of objects defined in the database: 129432
root 52936 69904 73 13:42:37 - 963:18 /usr/OV/bin/ovwdb -O
-n130000 -t
Very close! Looks like I should raise the cache size. How much higher
would you suggest? Our network recently went through some growth so this
explains what I am seeing. Would these number being so close cause the
slowness I am seeing now?
Thanks!
Scott
-----Original Message-----
From: owner-nv-l@lists.us.ibm.com [mailto:owner-nv-l@lists.us.ibm.com]
On Behalf Of Bursik, Scott {PBSG}
Sent: Thursday, June 01, 2006 11:06 AM
To: nv-l@lists.us.ibm.com
Subject: RE: [nv-l] nvUtil and nvcold
Thanks for the tips James. Yesterday nvUtil commands weren't coming back
at all and after the daemon refresh it is just returning slowly. I will
make some checks and see what I can find.
Scott
-----Original Message-----
From: owner-nv-l@lists.us.ibm.com [mailto:owner-nv-l@lists.us.ibm.com]
On Behalf Of James Shanks
Sent: Thursday, June 01, 2006 9:58 AM
To: nv-l@lists.us.ibm.com
Subject: Re: [nv-l] nvUtil and nvcold
I think you mean
NetView 7.1.4 AIX 5.2
There's a big difference between "never comes back" and "takes a long
time
to return". The latter indicates a performance problem, and to solve
that
you need to look at what else is going on in the system at the time and
what other things use smartsets. Got rulesets accessing smartsets too?
How about xnmsnmpconf? snmpCollect?
One other thing to remember. nvcold gets his information from ovwdb.
Smartsets, after all, are just objects in the database with various
attributes set. So anything that slows down ovwdb could slow down
nvcold.
Does ovobjprint work OK when this happens? Is you ovwdb cache size
larger
than the number of objects ovobjprint -S shows you? You wouldn't
necessarily have to add a new smartset to cause things to slow down. It
might be sufficient to add many new members to exisiting ones or
significantly add new objects to the database.
James Shanks
Level 3 Support for Tivoli NetView for UNIX and Windows
Tivoli Software / IBM Software Group
"Bursik, Scott
{PBSG}"
<Scott.Bursik@pbs
To
g.com> <nv-l@lists.us.ibm.com>
Sent by:
cc
owner-nv-l@lists.
us.ibm.com
Subject
[nv-l] nvUtil and nvcold
06/01/2006 10:45
AM
Please respond to
nv-l@lists.us.ibm
.com
NetView 6.1.4 AIX 5.2
I am having an issue where nvcold quits responding. It still shows as
running but when I try and run nvUtil commands it either never comes
back
or takes a long time to return. I recycled all of the daemons 2 times
yesterday and the issue keeps coming back. I haven't added any new
Smartsets. Any suggestions?
Thanks!
Scott Bursik
PepsiCo
|