RE: ovwdb startup option "-b"

["Smith, Kristi" <kristi_smith@mentorg.com>]

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

Can you relate some symptoms from the situation you mentioned:  "I had one
system run really badly after migrating from V5 to V6.01, but run fine after
I rediscovered."   We upgraded a couple of weeks ago and have intermittently
had problems with the ovwdb daemon hitting 50% CPU and staying there for
hours. While it's running at 50% CPU, the ICMP polling is being delayed so
node down notifications can be hours old. The SNMP queries seem to be
working fine. We shouldn't have any problem with the ovwdb cache or the
processing power of the server.  We end up having to bounce the daemons to
clear the problem.  

We are also having issues with NetView polling our 3Com LinkSwitches.  The
LinkSwitches come pre-configured with a SLIP address (192.101.1.1 or
192.168.101.1) and NetView 6.01 seems to be alot more sensitive to that
duplicate address everywhere (plus they're now viewed as routers too.) We
are getting rid of those addresses as we get downtime to reset the switches.

Thanks,
Kristi Smith
Mentor Graphics Corporation
(503) 685-1971
kristi_smith@mentor.com 


-----Original Message-----
From: Leslie Clark/Southfield/IBM [mailto:lclark@US.IBM.COM]
Sent: Wednesday, October 11, 2000 5:34 PM
To: IBM NetView Discussion
Subject: Re: [NV-L] ovwdb startup option "-b"


Peter, do you have the latest maintenance on Netview?
And does it idle out sometimes? I mean, what does the steady state look
like?
I  imagine that there is a vmstat command on Solaris. Is it paging?

That is a high object count, in my book. I would expect some daemons
to hog the system from time to time, even at 450 Mhz. But not all the time.

And is there any chance that there is corruption in the database? I usually
only see ovwdb working that hard when there is something wrong with
the database. You might consider calling Support. Or, if you can
take it out of production for half a day (or nigt), you might consider
backing
up the database and trying some things:
Try compression.
Try the nvTurboDatabase (speed) conversion (Read up on it first).
And try a fresh discovery (use a big seedfile made from your current
discovered base to speed things up) and see if it behaves any differently.
I had one system run really badly after migrating from V5 to V6.01, but run
 fine after I rediscovered.

Did it ever run well? Do you have a backup from that time that you can
restore and test? That would  tell you something.

I have absolutely no sway with Development. You will need to pursue
the -b idea via Support. I for one would be afraid to touch it, and accept
James' explanation that it was tuned to the best of their abilities.  Your
cache
setting is way higher than it needs to be, if your count is pretty steady.
Your
object count is not unusual. I'm pretty sure normal measures are called
for,
not something as arcane as -b. And those are the kind of measures I have
suggested above.

Keep us posted, eh?

Cordially,

Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit

"Peter Anderson" <pETERanderson@westpac.com.au>@tkg.com on 10/11/2000
06:53:01 PM

Please respond to IBM NetView Discussion <nv-l@tkg.com>

Sent by:  owner-nv-l@tkg.com


To:   "Netview List (E-mail)" <nv-l@tkg.com>
cc:
Subject:  Re: [NV-L] ovwdb startup option "-b"



Hi Leslie,

We have 24842 objects and a cache of 50000 running on a SUN Enterprise 450
(2x450Mhz,1GbRAM) ad it still runs slow as a dog at times.  ovwdb is quite
often the culprit - even if netmon is not running.  It will use 100% of one
CPU ad cause ovw_binary to stop processing I/O to the user for sometimes
more than 5 mins.

Can you please investigate this -b option further with the developers?

Thanks,

Peter Anderson
Senior Communications Analyst
Data Communications Support
Ph:  9902(5)5938




-----Original Message-----
From: SZEWCZYK, Jack
Sent: Wednesday, October 11, 2000 11:30 PM
To: ANDERSON, Peter
Subject: FW: Re: [NV-L] ovwdb startup option "-b"




-----Original Message-----
From:               Leslie Clark/Southfield/IBM
Sent:               Wednesday, 11 October 2000 12:27
To:            IBM NetView Discussion
Cc:
Subject:            Re: [NV-L] ovwdb startup option "-b"


-----Original Message-----
From:               Leslie Clark/Southfield/IBM
Sent:               Wednesday, 11 October 2000 12:27
To:            IBM NetView Discussion
Cc:
Subject:            Re: [NV-L] ovwdb startup option "-b"



Well, the commandline usage info does not match the man page for ovwdb.
I'll bet that option does not actually do anything any more. The one you
want
to mess with is the Cache. Make sure that is at least 20% greater than the
number of objects shown in the output of 'ovobjprint -S'. If cache is close
to
or less than that, and it defaults to 5000, you will indeed see super
slowness.

Cordially,

Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit

"Peter Anderson" <pETERanderson@westpac.com.au>@tkg.com on 10/11/2000
02:40:52 AM

Please respond to IBM NetView Discussion <nv-l@tkg.com>

Sent by:  owner-nv-l@tkg.com


To:   "Netview List (E-mail)" <nv-l@tkg.com>
cc:
Subject:  [NV-L] ovwdb startup option "-b"



Hi there,

I found a command line startup option for ovwdb which looks interesting to
try ...

However I can't find ant doco for it (so far ...)

The output from "ovwdb -h" reports:

Usage: ovwdb [-O] [-d dbname] [-P <port>] [-n <number_of_objects>] [-b
db_buffer_size]
        -O                     Assume started by ovspmd
        -d dbname              Set name of database
        -P port                Use <port> as the TCP port number
        -n number_of_objects   Cache only a limited number of objects
        -b db_buffer_size      Use a temporary <buffer_size> byte buffer in
the database to increase performance

I'm interested in trying the -b option as our system runs REAL slow.

nvsetup doesn't let you change most of these options, so I'd have to edit
the lrf file.

Anyone know if there is a default size setting, or is there none at all?

We have Netview Unix 6.01

Regards,

Peter Anderson
Senior Communications Analyst
Westpac Banking Corporation
Ph:  0011 61 2 99025938
<Remove ETER from my address to reply>

Any views or opinions presented are solely those of the author and do not
necessarily represent those of Westpac Banking Corporation.

_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l


_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l

_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l


_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l

Can you relate some symptoms from the situation you mentioned:  "I had one system run really badly after migrating from V5 to V6.01, but run fine after I rediscovered."   We upgraded a couple of weeks ago and have intermittently had problems with the ovwdb daemon hitting 50% CPU and staying there for hours. While it's running at 50% CPU, the ICMP polling is being delayed so node down notifications can be hours old. The SNMP queries seem to be working fine. We shouldn't have any problem with the ovwdb cache or the processing power of the server.  We end up having to bounce the daemons to clear the problem. 

We are also having issues with NetView polling our 3Com LinkSwitches.  The LinkSwitches come pre-configured with a SLIP address (192.101.1.1 or 192.168.101.1) and NetView 6.01 seems to be alot more sensitive to that duplicate address everywhere (plus they're now viewed as routers too.) We are getting rid of those addresses as we get downtime to reset the switches.

Thanks,
Kristi Smith
Mentor Graphics Corporation
(503) 685-1971
kristi_smith@mentor.com

-----Original Message-----
From: Leslie Clark/Southfield/IBM [mailto:lclark@US.IBM.COM]
Sent: Wednesday, October 11, 2000 5:34 PM
To: IBM NetView Discussion
Subject: Re: [NV-L] ovwdb startup option "-b"

Peter, do you have the latest maintenance on Netview?
And does it idle out sometimes? I mean, what does the steady state look
like?
I  imagine that there is a vmstat command on Solaris. Is it paging?

That is a high object count, in my book. I would expect some daemons
to hog the system from time to time, even at 450 Mhz. But not all the time.

And is there any chance that there is corruption in the database? I usually
only see ovwdb working that hard when there is something wrong with
the database. You might consider calling Support. Or, if you can
take it out of production for half a day (or nigt), you might consider
backing
up the database and trying some things:
Try compression.
Try the nvTurboDatabase (speed) conversion (Read up on it first).
And try a fresh discovery (use a big seedfile made from your current
discovered base to speed things up) and see if it behaves any differently.
I had one system run really badly after migrating from V5 to V6.01, but run
 fine after I rediscovered.

Did it ever run well? Do you have a backup from that time that you can
restore and test? That would  tell you something.

I have absolutely no sway with Development. You will need to pursue
the -b idea via Support. I for one would be afraid to touch it, and accept
James' explanation that it was tuned to the best of their abilities.  Your
cache
setting is way higher than it needs to be, if your count is pretty steady.
Your
object count is not unusual. I'm pretty sure normal measures are called
for,
not something as arcane as -b. And those are the kind of measures I have
suggested above.

Keep us posted, eh?

Cordially,

Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit

"Peter Anderson" <pETERanderson@westpac.com.au>@tkg.com on 10/11/2000
06:53:01 PM

Please respond to IBM NetView Discussion <nv-l@tkg.com>

Sent by:  owner-nv-l@tkg.com

To:   "Netview List (E-mail)" <nv-l@tkg.com>
cc:
Subject:  Re: [NV-L] ovwdb startup option "-b"

Hi Leslie,

We have 24842 objects and a cache of 50000 running on a SUN Enterprise 450
(2x450Mhz,1GbRAM) ad it still runs slow as a dog at times.  ovwdb is quite
often the culprit - even if netmon is not running.  It will use 100% of one
CPU ad cause ovw_binary to stop processing I/O to the user for sometimes
more than 5 mins.

Can you please investigate this -b option further with the developers?

Thanks,

Peter Anderson
Senior Communications Analyst
Data Communications Support
Ph:  9902(5)5938

-----Original Message-----
From: SZEWCZYK, Jack
Sent: Wednesday, October 11, 2000 11:30 PM
To: ANDERSON, Peter
Subject: FW: Re: [NV-L] ovwdb startup option "-b"

-----Original Message-----
From:               Leslie Clark/Southfield/IBM
Sent:               Wednesday, 11 October 2000 12:27
To:            IBM NetView Discussion
Cc:
Subject:            Re: [NV-L] ovwdb startup option "-b"

-----Original Message-----
From:               Leslie Clark/Southfield/IBM
Sent:               Wednesday, 11 October 2000 12:27
To:            IBM NetView Discussion
Cc:
Subject:            Re: [NV-L] ovwdb startup option "-b"

Well, the commandline usage info does not match the man page for ovwdb.
I'll bet that option does not actually do anything any more. The one you
want
to mess with is the Cache. Make sure that is at least 20% greater than the
number of objects shown in the output of 'ovobjprint -S'. If cache is close
to
or less than that, and it defaults to 5000, you will indeed see super
slowness.

Cordially,

Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit

"Peter Anderson" <pETERanderson@westpac.com.au>@tkg.com on 10/11/2000
02:40:52 AM

Please respond to IBM NetView Discussion <nv-l@tkg.com>

Sent by:  owner-nv-l@tkg.com

To:   "Netview List (E-mail)" <nv-l@tkg.com>
cc:
Subject:  [NV-L] ovwdb startup option "-b"

Hi there,

I found a command line startup option for ovwdb which looks interesting to
try ...

However I can't find ant doco for it (so far ...)

The output from "ovwdb -h" reports:

Usage: ovwdb [-O] [-d dbname] [-P <port>] [-n <number_of_objects>] [-b
db_buffer_size]
        -O                     Assume started by ovspmd
        -d dbname              Set name of database
        -P port                Use <port> as the TCP port number
        -n number_of_objects   Cache only a limited number of objects
        -b db_buffer_size      Use a temporary <buffer_size> byte buffer in
the database to increase performance

I'm interested in trying the -b option as our system runs REAL slow.

nvsetup doesn't let you change most of these options, so I'd have to edit
the lrf file.

Anyone know if there is a default size setting, or is there none at all?

We have Netview Unix 6.01

Regards,

Peter Anderson
Senior Communications Analyst
Westpac Banking Corporation
Ph:  0011 61 2 99025938
<Remove ETER from my address to reply>

Any views or opinions presented are solely those of the author and do not
necessarily represent those of Westpac Banking Corporation.

_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l

_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l

_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l

_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l