nv-l
[Top] [All Lists]

Re: [nv-l] ovactiond problem

To: nv-l@lists.us.ibm.com
Subject: Re: [nv-l] ovactiond problem
From: James Shanks <jshanks@us.ibm.com>
Date: Thu, 9 Mar 2006 12:50:26 -0500
Delivery-date: Thu, 09 Mar 2006 17:50:49 +0000
Envelope-to: nv-l-archive@lists.skills-1st.co.uk
In-reply-to: <4253B9BA00058A14@pop3.es.tisadm.net>
Reply-to: nv-l@lists.us.ibm.com
Sender: owner-nv-l@lists.us.ibm.com
Is there some reason why you would choose this method of the internal TEC
adapter in nvserverd?   That method is optimized by having nvcorrd pick out
the correct events to send and then having nvserverd do the formatting and
sending (two daemons rather than one share the processing).

Using a ruleset and the internal adapter is far more efficient than using
wpostemsg from a script if you are going to send a lot of events.
But I don't see how setting the -w to zero could hurt you, since the TEC
server can handle out-of-sequence events if that becomes an issue.  The
thing is that once the scripts are launched, you have no control of which
one(s) the operating system gives priority to.  Usually it's FIFO but you
never know.

Anybody else sending events to TEC with using postemsg or wpostemsg have
something they'd like to share?

James Shanks
Level 3 Support  for Tivoli NetView for UNIX and Windows
Tivoli Software / IBM Software Group


                                                                           
             "Gerardo Ruiz                                                 
             Garcia"                                                       
             <gerardo_ruiz@ini                                          To 
             cia.es>                   nv-l@lists.us.ibm.com               
             Sent by:                                                   cc 
             owner-nv-l@lists.                                             
             us.ibm.com                                            Subject 
                                       Re: [nv-l] ovactiond problem        
                                                                           
             03/09/2006 12:36                                              
             PM                                                            
                                                                           
                                                                           
             Please respond to                                             
             nv-l@lists.us.ibm                                             
                   .com                                                    
                                                                           
                                                                           




The design is tio execute an script for each event that means a TEC message
executng a wpostemsg comand in the script.
In this way we can create our own class definition
The mayor traps are received from Cisco Router snmp agent: interface limks,
environment problems etc..
Only a few tarps or events are configured to send wpostemsg to TEC, but
the rate of traps/events makes ovactiond disconect from trapd.
I do not understand the meaning of serializing a resource.

Thanks for the comments I will try to set -W to cero and inncrease size
of trapd.
What is your opinion of the desing for TEC connection.?


>-- Mensaje Original --
>Subject: Re: [nv-l] ovactiond problem
>To: nv-l@lists.us.ibm.com
>From: James Shanks <jshanks@us.ibm.com>
>Date: Thu, 9 Mar 2006 08:03:02 -0500
>Reply-To: nv-l@lists.us.ibm.com
>
>
>Probably so, especially if you are serializing on some resource.  What
is
>it that you are trying to do?
>
>James Shanks
>Level 3 Support  for Tivoli NetView for UNIX and Windows
>Tivoli Software / IBM Software Group
>
>
>

>
>             "Gerardo Ruiz

>
>             Garcia"

>
>             <gerardo_ruiz@ini
 To
>
>             cia.es>                   nv-l@lists.us.ibm.com

>
>             Sent by:
 cc
>
>             owner-nv-l@lists.

>
>             us.ibm.com                                            Subject
>
>                                       Re: [nv-l] ovactiond problem

>
>

>
>             03/09/2006 04:53

>
>             AM

>
>

>
>

>
>             Please respond to

>
>             nv-l@lists.us.ibm

>
>                   .com

>
>

>
>

>
>
>
>
>
>I have a question to the option wait -w set to cero.
>THat it means that the script are executed in background? , so I have to
>develop the scripts in a way that several instances of the script can we
>executed at the same time.
>
>>-- Mensaje Original --
>>Subject: Re: [nv-l] ovactiond problem
>>To: nv-l@lists.us.ibm.com
>>From: James Shanks <jshanks@us.ibm.com>
>>Date: Wed, 8 Mar 2006 09:14:26 -0500
>>Reply-To: nv-l@lists.us.ibm.com
>>
>>
>>Francois is probably right that increasing the application queue size
will
>>help resolve your problem, but you should be aware of the performance
>>implications of what you are doing.  The reason that ovactiond is losing
>>contact with trapd is that he cannot execute whatever it is that you are
>>having him execute fast enough to keep up.  So his trap queue backs up
>>inside trapd  and then trapd closes it when it gets full.  trapd can
>handle
>>receiving 10 traps a second and passing them on, but ovactiond cannot
take
>>action on every one at that rate.  So making the queue size bigger will
>>probably allow ovactiond to stay connected, but he will fall behind.
The
>>only way he can catch up is if the trap rate to him slows down
eventually,
>>or if he doesn't wait for the actions you ask him to execute to complete.
>>By default they are serialized.  But you can have him just spawn them
as
>>fast as he can by setting the wait time (-w) to zero.
>>See the man page on ovactiond.
>>
>>The entire idea behind trap processing is that it should be "bursty",
that
>>is, that there should be slow periods as well as fast ones.  If there
>>aren't any slow periods, then your daemons are likely to fall behind,
>>depending on how much work you have them do.     I don't have any
>>benchmarks to share with you, so you are going to have to watch your
>system
>>for yourself.
>>
>>James Shanks
>>Level 3 Support  for Tivoli NetView for UNIX and Windows
>>Tivoli Software / IBM Software Group
>>
>>
>>
>
>>
>>             Francois Le Hir
>
>>
>>             <flehir@ca.ibm.co
>
>>
>>             m>
> To
>>
>>             Sent by:                  nv-l@lists.us.ibm.com
>
>>
>>             owner-nv-l@lists.
> cc
>>
>>             us.ibm.com
>
>>
>>
Subject
>>
>>                                       Re: [nv-l] ovactiond problem
>
>>
>>             03/08/2006 08:18
>
>>
>>             AM
>
>>
>>
>
>>
>>
>
>>
>>             Please respond to
>
>>
>>             nv-l@lists.us.ibm
>
>>
>>                   .com
>
>>
>>
>
>>
>>
>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>you can increase the value of "Set Trapd connected applications queue
>size"
>>in the trapd configuration (smitty nv6000 or serversetup). I don't
>remember
>>the default value but mine is now set to 25000 (which is probably much
>more
>>than the default).
>>
>>Salutations, / Regards,
>>
>>Francois Le Hir
>>Network Projects & Consulting Services
>>IBM Global Services
>>
>>
>>
>>
>>             "Gerardo Ruiz
>>             Garcia"
>>             <gerardo_ruiz@ini
> To
>>             cia.es>                   nv-l@lists.us.ibm.com
>>             Sent by:
> cc
>>             owner-nv-l@lists.
>>             us.ibm.com
Subject
>>                                       [nv-l] ovactiond problem
>>
>>             03/08/2006 04:34
>>             AM
>>
>>
>>             Please respond to
>>             nv-l@lists.us.ibm
>>                   .com
>>
>>
>>
>>
>>
>>
>>Hi list:
>>In our Netview design it is executed a script with several traps are
>>received,
>>in trapd configuracion.
>>As the rate of traps is 10 per second the ovactiond has problems, with
>the
>>following message in ovactiond.log:
>>
>>03/06/06 10:00:33 ovactiond lost contact with trapd.
>>03/06/06 10:03:40 ovactiond lost contact with trapd.
>>03/06/06 10:14:46 ovactiond lost contact with trapd.
>>03/06/06 10:29:16 ovactiond lost contact with trapd.
>>03/06/06 10:31:14 ovactiond lost contact with trapd.
>>
>>How can I improbe behaviour of ovactiond.
>>trapd is configured with default settings.
>>Netview 7.1.4 FP03 AIX5.2 ML-6 .
>>
>>Any comments wil be appreciated.
>>
>>
>>
>>
>>¿Quieres sacar el máximo a tu conexión Tiscali?
>>
>>Toda la información sobre como configurar tu conexión
>>en nuestra sección de Ayuda.
>>http://ayuda.tiscali.es/ayuda
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>
>
>¿Quieres sacar el máximo a tu conexión Tiscali?
>
>Toda la información sobre como configurar tu conexión
>en nuestra sección de Ayuda.
>http://ayuda.tiscali.es/ayuda
>
>
>
>
>
>
>


¿Quieres sacar el máximo a tu conexión Tiscali?

Toda la información sobre como configurar tu conexión
en nuestra sección de Ayuda.
http://ayuda.tiscali.es/ayuda








<Prev in Thread] Current Thread [Next in Thread>

Archive operated by Skills 1st Ltd

See also: The NetView Web