Finally, I found a race condition inside my access code. So, master-subagent on multiple walks should be ok.
Thanks, FT -----Original Message----- From: Magnus Fromreide [mailto:[EMAIL PROTECTED] Sent: Wednesday, June 01, 2005 10:44 PM To: Fong Tsui Cc: [email protected] Subject: Re: performance issue on simultaneously walk On Wed, Jun 01, 2005 at 05:35:59PM -0700, Fong Tsui wrote: > > I have a subagent behind a master agent. When I do a single walk a MIB > tree, it is very fast. However, when I do simultaneously walk on the > same MIB tree, both walk take unbelievable long time. One report says > 5min vs 45 sec (two walk vs single walk). > > I suspect two things: > 1) That master agent talking to subagent with simultaneous queries slows > down the speed? That shouldn't be the the case - there is additional overhead as you are issuing twice as many requests but it shouldn't be that bad and as the times you are quoting are very big I'd say you have either a very large number of objects or there are some other problem. > 2) My code for accessing our system data doesn't support multiple tasks > well? I suppose this is the case but I also suspect that you could speed things up in the single walk case. If you build your subagent as part of the agent rather than as a subagent, how does that affect the times involved? This would be a easy test of the theory. > Does anybody see 1) could be a problem? If yes, could it make so much > difference? > /MF ------------------------------------------------------- This SF.Net email is sponsored by: NEC IT Guy Games. How far can you shotput a projector? How fast can you ride your desk chair down the office luge track? If you want to score the big prize, get to know the little guy. Play to win an NEC 61" plasma display: http://www.necitguy.com/?r _______________________________________________ Net-snmp-coders mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
