Thanks Dave for the reply.
Making AgentXRetries to 0 avoids sending multiple copies from Agentx. It really
help to reduce the impact.
But the problem still exists in snmp response which I earlier thought this was
because of multiple Agentx Retries.
I sent multiple SNMP queries (packed 5 in a single request) multiple times (in
a loop).
I am getting 'no such instance' during heavy load from my app that response to
query.
Diagramatically During heavy load
---------------------------------
Client Agent AgentX myAPP
------ ----- ------ -----
1. Sends 1st request ----------->
Agentx_recved
|----------------->
myApp handler trigd
|---------------------->
myApp responded
<-----------------------|
(Before that client timeout happens)
2. Client resending request------>
Agentx_recved
|----------------->
AgentX sends response (1)
<------------------|
(No Error in AgentX packet structure)
errStatus - 0, ErrIndex -0)
Packet with GenErr and Err-index(1)
<-------------------|
(First packet)
triggers second get
|-------------------->
The problem here that I am facing now is that, Agentx gets the value from my
app and sends the same to Agent (dumpx_send calls shows the same without
error). But the actual response sent to client contains the gen-err and
error-index-1 with no such object as the value.
This causes the client (snmpclient) to resend the packet (with one varbind
less) and which in turn cause 'no such instance object' for the next 20 set of
SNMPget packet request.
Communication between Agentx and myAPP works fine. But the resultant response
to client contains 'no such object' value.
Any pointers to this.
Is there any way to get the log of agent (agent communication to client).
Thanks in Advance for your help
Prakash.V
--- On Fri, 4/23/10, Dave Shield <[email protected]> wrote:
> From: Dave Shield <[email protected]>
> Subject: Re: Agentx receiving multiple SNMP requrest though few were sent
> To: "surya prakash" <[email protected]>
> Cc: "net-snmp-coders" <[email protected]>
> Date: Friday, April 23, 2010, 9:22 PM
> On 19 April 2010 12:59, surya prakash
> <[email protected]>
> wrote:
> > If I increase the load on my server (machine which
> runs the app that registers the handler),
> > the behavior changes. I enabled log to analyze the
> issue.
> >
> > On Analyzing, I found that AgentX receives multiple
> request for the same OID though a
> > single snmpget is called. I am running a sniffer in
> between which tells that only one
> > snmpget was sent to the server. But AgentX receives
> multiple copies and my app
> > response for the same.
>
> Have you tried experimenting with the 'AgentXTimeout' and
> 'AgentXRetries' parameters?
> See snmpd.conf(5)
>
> Dave
>
------------------------------------------------------------------------------
_______________________________________________
Net-snmp-coders mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
