[AGENT++] Timeout seen for GETNEXT when an object is excluded from a view

[Maurice Plunkett]

Thanks Frank,

Yes, I agree that it is not a trivial fix by any means.  Thank you for your
attention to this matter.

Best regards,
	Maurice.


 <http://www.efi.com/>    
             
Maurice Plunkett
Lead Software Engineer

303 Velocity Way
Foster City, CA 94404
[650] 357 3875 tel
[650] 357 3994 fax
maurice.plunkett at efi.com


-----Original Message-----
From: Frank Fock [mailto:fock at agentpp.com] 
Sent: Tuesday, June 21, 2005 9:50 PM
To: Maurice Plunkett
Cc: agentpp at agentpp.org
Subject: Re: [AGENT++] Timeout seen for GETNEXT when an object is excluded
from a view

Hi Maurice,

Sure the algorithm can be optimized. However, it is not that easy
as it might seem, because the VACM may include a subtree within
an excluded region. I will try to provide a solution in one of the
next releases...

Best regards,
Frank


Maurice Plunkett wrote:

>Hi,
>
> 
>
>Is it possible to make GETNEXT more efficient in the case when the target
>object is excluded from a user's view?
>
> 
>
>For example,
>
>If ifTable (1.3.6.1.2.1.2.2) is excluded from view via an entry in the
>vacmViewTreeFamilyTable, a GETNEXT request to ifNumber.0 results in the
agent
>effectively walking the ifTable until it finds the next object after the
>ifTable.
>
> 
>
>This may take a relatively short period of time for the ifTable, so the
>client may not time-out.
>
> 
>
>However, if larger MIB trees are excluded from a view, such as the Host
>Resource MIB or an enterprise MIB, this delay can be too long.  It may also
>appear to a client that the agent has died.  It would be best if the client
>did not experience such long delays.
>
> 
>
>Perhaps one work-around would be for the agent to do a GET operation on the
>object that is lexicographically after the excluded view and start from
>there...
>
> 
>
>Regards,
>
>            Maurice.
>
> 
>
> 
>
>_______________________________________________
>AGENTPP mailing list
>AGENTPP at agentpp.org
>http://lists.agentpp.org/mailman/listinfo/agentpp
>
>  
>