[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: First httpd deadlock fixed, but found another one

From: Justin Erenkrantz <jerenkrantz_at_apache.org>
Date: 2002-12-11 04:03:53 CET

--On Tuesday, December 10, 2002 5:13 PM -0500 Brandon Ehle
<azverkan@yahoo.com> wrote:

> Hrmm, are you sure? I see it try to shutdown, then if that doesn't
> work it starts sending SIGTERMs to processes. If the processes
> ignore the SIGTERM's, it will wait for them to finish which could
> be hours or never, unless I manually send a SIGKILL, which does the
> job, but the repository usually needs recovery at that point.

You are correct. The worker MPM will wait infinitely for its
children to come back. If it never exits (due to an infinite loop in
BDB), then that could be a problem.

Perhaps sending an email to the dev@httpd list asking about this
would be beneficial. Not sure how you would do this since the
pthread_join call is usually blocking. Perhaps doing a pthread_kill
with SIGKILL beforehand might be helpful. Anyway, dev@svn isn't the
right forum for this part of the bug. -- justin

To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Dec 11 04:04:34 2002

This is an archived mail posted to the Subversion Dev mailing list.