[ic] workaround found for 99%CPU usage problem in Vend/SessionFile.pm: "::uneval($ref);" fixes it

Dan Browning interchange-users@interchange.redhat.com
Wed Nov 14 19:43:00 2001


> * Mike Heins wrote on Wed, Nov 14, 2001 at 15:39 -0500:
> > Quoting Steffen Dettmer (steffen@dett.de):
>  [...] 
> > > And it works, too. No 99% CPU processes. I don't know what unveal 
> > > does, but it workarounds the problem. I'm afraid that this don't 
> > > fixes the issue but gives a point less where it gets triggered.
> > > 
> > > For completness a patch:
> > 
> > I don't see any context here, but I do recommend that no one do this
> 
> Hum, yep, of course, I forgot to say it explicitly that this 
> patch is not for any user but for developer only, and only to 
> make clear what I changed exactly. Second, this patch is 
> empiric guessed as I tried to insinuate.
> 
> What do you mean with "I don't see any context here"? To less 
> patch context? I just used diff -u which defaults to 3 lines context.
> 
> > I am guessing your operating system is BSD.
> 
> I'm running linux.
> 
> > Why I should have to guess, I don't know -- it should be 
> mandatory to 
> > mention that in a message like this.
> 
> You're right, sorry. It's an SuSE 7.0 distribution:
> Linux stella 2.2.16-SMP #1 SMP Wed Aug 2 20:01:21 GMT 2000 
> i686 unknown with the old Perl 5.005_03.
> 
> > In any case, the thing you mentioned makes it seem like you have a 
> > problem with file locks. I would run my "make test" on Perl and see 
> > whether it passed 100%.
> 
> Thank you for the hint. You suggest to do a make test in the 
> perl sources, right? I'll have to upgrade Perl anyway, so I'll do it.
> 
> File locks could really be problematic, since the file system 
> is a NFS share. So this may be a NFS problem? I could check 
> it by moving the home of this user to a local file system, 
> this may be a nice idea, correct? 

For NFS, try this in interchange.cfg:

LockType fcntl

I thought that was in the docs?  Hmm.

Dan Browning
Kavod Technologies