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

Re: 500 Internal Server Error: corrupted BDB

From: <Matthew_Rich_at_playstation.sony.com>
Date: 2004-01-30 21:45:52 CET

Well, it's an old 1.8, so no hyper-threading, fortunately or
unfortunately as the case may be ;) It seems like any sufficiently
massive dataset would do the trick. I don't think it's specific to my
data, in any case I've nuked the repository and tried it from scratch
a couple of times with the same result.
I wouldn't be using Fedora with an SMP kernel in any case, unless
they've fixed it.

1)
I'm using a basic parent-path setup with authentication, SSL ...
suppose I should also run a test without those variables in the mix.

<Location /source>
DAV svn
SVNParentPath /home/svnroot/
 
AuthName "source"
AuthType Basic

...
 
require valid-user
AllowOverride None
Order allow,deny
Allow from all
</Location>

2)
svnadmin create /home/svnroot/gcc
chown -R apache.apache /home/svnroot/gcc

3) I did an add/commit rather than an import because I was originally
working with an existing repository.

svn co https://anduril/source/gcc
mv gcc-3.2 gcc
cd gcc
svn add gcc-3.2
svn commit -m ''

I'm in the process of running a test with a simplified configuration:
just plain HTTP, no auth.
Thanks.

Cheers,

Matt

On Fri, 2004-01-30 at 02:32, John Szakmeister wrote:
> On Thursday 29 January 2004 23:02, Matthew_Rich@playstation.sony.com wrote:
> > Interesting, so 0.37.0 through mod_dav_svn still timeouts on commit, but
> > it also consistently seems to cause httpd to become unresponsive, and
> > saturate the cpu. These problems don't seem to
> > happen when accessing the repository locally (file:///...). I seem to
> > recall reading that it was possible to increase the timeout for neon;
> > haven't tried that yet but that still wouldn't explain the strange
> > behaviour with Apache. fwiw, I still need to go through and test with BDB
> > 4.2, but there again it doesn't seem to be a problem with the DB since it
> > worked correctly for a direct repository access.
>
> You're running Fedora, which has 4.1.25, right? I also see that you have a P4
> 1.8GHz machine... does it happen to have hyper-threading (does it show up as
> a dual-cpu machine)? If so, I can tell you now that I've been keeping an eye
> on the correlation of problems to 4.1.25, and it's pretty high when it comes
> to multi-processor machines. While a single P4 w/HT may not be a true
> multi-processor machine, it's close enough. I'd definitely give 4.2.x a go.
>
> Anyway you can see where it's hanging in the server process when this happens?
> Actually, if you can tell me what you were doing, I'll try on my machine to
> see if I can make it happen, and then we can do some bug hunting together.
> Were you simply importing the gcc code?
>
> -John

-- 
Matthew Rich <Matthew_Rich@playstation.sony.com>
Sony Computer Entertainment America
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Jan 30 21:46:10 2004

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