over 200 threads running - solved!

[Stephen S. Kelley]

I forgot to change the server names in some places (in particular
[config.root].mainresponser.search.domain) after moving the software.
Index requests were piling up in user.betty.queueOutgoing to the old
server, timing out then getting re-queued for 5 minutes later.

 

________________________________

From: frontier-users-bounces at userland.com
[mailto:frontier-users-bounces at userland.com] On Behalf Of Steve Hooker
Sent: Sunday, September 26, 2004 4:07 AM
To: Frontier-Users at userland.com
Subject: Re: over 200 threads running - more info

 

What and where will be difficult to guess from a distance. You may want
to dig into the Mainresponder.root threads too at
mainResponder.background.everyMinute. But strange you say every few
minutes not every minute.

Perhaps you may think about your DNS and tcp.httpclient. There's a line
in that script that goes something like this, "Turn our worst case into
our best case." Perhaps 'something' like the aggregator is trying to get
local feeds and failing. Whatever it is, each thread, because of some
DNS mistake will take an extremely long time to time out. Hope this
helps

Steve Hooker
http://www.cybersaps.org/
 


Stephen S. Kelley wrote: 

The crash was only running thread.getStat from quickscript, here are
what all the threads look like:

 

{{191, 57, "system.verbs.builtins.tcp.openStream"}, {191, 58,
"system.verbs.builtins.tcp.httpClient"}, {119, 224,
"system.verbs.builtins.betty.rpc.client"}, {14, 219,
"system.verbs.builtins.betty.rpc.sendOneMessage"}, {1, 73, "anonymous"}}

 

I put a return at the beginning of
system.verbs.builtins.betty.rpc.sendOneMessage until I can figure out
the real cause.

 

________________________________

From: frontier-users-bounces at userland.com
[mailto:frontier-users-bounces at userland.com] On Behalf Of Stephen S.
Kelley
Sent: Saturday, September 25, 2004 7:12 PM
To: frontier-users at userland.com
Subject: over 200 threads running

 

Since I've moved frontier 9.0 over to windows 2003 it periodically
(every few minutes) opening lots of threads, frequently over 200 which
eventually go away and down to 3-5 which is usual. I've tried changing
the time for everyminute scheduler tasks to a large value to deduce that
it is not something getting kicked off by the scheduler.

 

I've also tried running thread.getStats to find information about the
threads, but this crashes frontier and I don't get anything.

 

Help!

 

Dr. Stephen S. Kelley, President
VirtuState, Inc.
stevek at virtustate.com
http://www.virtustate.com <http://www.virtustate.com/> 

 






-- 
Steve Hooker
http://www.cybersaps.org
In  UK: 0800 849 6413
Out UK: +44 (0) 952 271 671
Mobile: +44 (0) 7903 940 427 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.userland.com/pipermail/frontier-users/attachments/20040926/be2f9458/attachment.htm