Re: Cricket crashing on network timeout

Hi Ed

The collector process quits competly, leaving the lock file.

I have managed to fix the problem, but still not 100% sure why my fix
worked. What I did was disabled all logging output by a quick and dirty perl
hack in the Log.pm file, which just returned from sub Log, rather than doing
the logging.

My guess is I have a misconfiguration somewhere, and when a device errored
it caused a broken pipe somewhere, I assume to a mail program?


email-program       = /usr/bin/mailx

Which doesn't exist, hence the broken pipe and the crash.

I am going to play around with it a bit more, but pretty sure I am heading
int the right direction.


-----Original Message-----
From: Ed Ravin [mailto:[email protected]] 
Sent: Saturday, 7 October 2006 3:41 a.m.
To: Simon Allard
Cc: [email protected]
Subject: Re: [cricket-users] Cricket crashing on network timeout

> I am using cricket 1.0.5 on Ubuntu edgy. It works really well but when I
> have a network event which causes the SNMP to timeout, the polling dies,
> which means the other appliances don?t get graphed until I notice and
delete
> the lock file, which really sucks when trying to debug why your network is
> dead :P

Please clarify what you mean by "the polling dies" - does the collector
process get stuck but stay running, or has it quit completely?

Did you find the output of the cron job that crashed?  What was in it?

> exactly the same problem with Ubuntu dapper, hence why I upgraded to edgy
to
> see if an upgrade in the SNMP libs would fix the problem.

Note that by default, Cricket doesn't use Net-SNMP libraries - you may
be barking up the wrong tree there.

Can you reproduce the problem, say, by trying to poll a device that
doesn't exist, or poll an existing device with the wrong community name?

> Also, interesting to note, that if I run cricket by hand in a shell, it
> works fine. 

If that's the case, and you can reproduce the problem from your cron job,
something is wrong with the permissions, environment (like the command path
or library path), or user ID (are your successful tests as the same user
that runs the job in cron?).

	-- Ed


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV