Re: beacon 0.9.58 dies after a while - broken pipe?
Hi,
>Date: Tue, 10 Aug 2004 09:58:08 -0500
>From: John Kristoff <jtk@northwestern.edu>
>To: beacon@dast.nlanr.net
>Subject: Re: beacon 0.9.58 dies after a while - broken pipe?
>
>On Mon, 09 Aug 2004 20:36:28 -0500
>Mitch Kutzko <mitch@ncsa.uiuc.edu> wrote:
>
>> Hi, John -- Added this change (and others) into today's release.
>
>Hmm... I'm not so sure that I would have recommended you apply that to
>the released version. It was just my way of getting around some other
>problem I've not yet uncovered.
Agreed, but it doesn't hurt anything to give it out as part of the release.
Just a mini-kludge to keep the Beacon from falling over so often until we
can figure out what's really wrong.
> No luck in determining if there is a
>problem at the central server? When I get a chance I'll see if I can
>spot an anomaly with a packet capture.
No luck so far -- Any help appreciated.
>The other way to hack around this problem might be run the beacon code
>in a startup script's while loop, if it exits, the while loop will start
>it up again. This throws off the uptime data, but maybe that would be
>a preferred method for some and maybe there is something that can be
>correlated at the server end by you knowing exactly when each client
>restarted?
Possibly, but I'd prefer to find what's actually wrong and fix it,
obviously. If between us we can't get it sorted out in the next week or
so, I'll see about adding something like this until we can.
>> Let me know it works for you -- I've received one report already saying
>> there are still problems with it.
>
>When you say there is still a problem, do you mean that someone is
>still seeing the script exit without warning or that there is a
>problem in that code that I had posted?
Still seeing the broken pipe causing an exit without warning. (I should
have been clearer about that.)
Thanks for your help!
Mitch
--
Mitch Kutzko | mitch@dast.nlanr.net | mitch@ncsa.uiuc.edu | 217-333-1199
http://hobbes.ncsa.uiuc.edu/