Re: {bb} System Hangs

[Charles Jones <jonescr@cisco.com>]

I was once having the exact same problem, and I used yum to install the
latest greatest stable kernels whenever they came out, but the problem
persisted.  I upgraded the box to FC1 and the problem went away.

P.S. The symptoms/error was that BB server process would start to
zombie and the box would become progressively more and more sluggish
(memory leak) until it locked up.

------------------ Snip /var/log/messages -----------------

May 24
23:15:37 BB_lnx kernel: application bug: bbd(30461) has SIGCHLD set to
SIG_IGN but calls wait().
May 24 23:15:37 BB_lnx kernel: (see the NOTES section of 'man 2 wait').
Workaround activated.

-Charles

Andrew Davis wrote:
> If that
> is the case, I don't think there is a "enable or disable X" answer.
> RedHat is known for customizing and modularizing their kernels, as well
> as backports and all sorts of other goodies (not that other distros
> don't do the same, RH just does more of it). If recompiling the kernel
> is the answer, get it kernel.org, do an audit of your devices and
> compile a custom kernel with just what you need to support your
> hardware. RH never ships the latest, greatest kernel anyway, so if
> there's a bug in your existing one, it'll likely be fixed in the
> latest, greatest one available at kernel.org.
>   
>   
> At this point, I must confess that I don't remember what the original
> complaint was. Perhaps some log or error file output would help us help
> you...
>   
>   
> AD
>   
>   
> Charles Jones wrote:
>   
>   
> > I think he meant to define "rebuilding the
> > kernel".  As in...if you simply compile your own kernel and choose all
> > the default options, does that fix the problem? I think the response he
> > is looking for is to recompile his kernel and to enable or disable
> > option X which causes the problem.
> >     
> >     
> > -Charles
> >     
> >     
> > Philip Clark wrote:
> >     
> >     
> > > On Thu, 2004-05-27 at 00:21, Wray Cason
> > > wrote:
> > >       
> > >  
> > >       
> > >       
> > > > This is not a very fulfilling answer. 
> > > > FC2 has an entirely different kernel
> > > >         
> > > > then RH9 or RH7.3.  You say to rebuild the kernel, do you mean rebuild
> > > > the
> > > >         
> > > > stock kernel of the vanilla kernel?  I doubt that any problem BB has
> > > > with RH9
> > > >         
> > > > has any relevance to BB on FC2.
> > > >         
> > > >    
> > > >       
> > >       
> > > The original poster sent a follow-up to qualify this. It is Red Hat 9
> > >       
> > > that he is having problems with.
> > >       
> > >       
> > >  
> > >       
> > >       
> > > > By the way, I have had problems with a
> > > > couple of my BB systems running RH9. It seems to be memory related in
> > > > some fashion since none of my machines with
> > > >         
> > > > at least 256mb ram are affected.  These systems have ram <= 128mb. 
> > > > These
> > > >         
> > > > systems sometimes experience increasing cpu load averages and the hard
> > > > drive
> > > >         
> > > > goes nuts with activity.  It gets progressively worse until I reboot. 
> > > > It can
> > > >         
> > > > get so bad that sshd can't respond.  Anyway, that is my interest in
> > > > this
> > > >         
> > > > discussion.
> > > >         
> > > >    
> > > >       
> > >       
> > > The list archives are the best place to look. IIRC the closest anyone
> > >       
> > > got to the cause were some "bleeding edge" changes to signal handling
> > >       
> > > that Red Hat saw fit to include. Some folks managed to get by using
> > >       
> > > the ZOMBIE flag when compiling BB. Building your own kernel appears
> > >       
> > > to be the most effective solution however.
> > >       
> > >       
> > > Cheers, Phil.
> > >       
> > >       
> > >       
> > >  
> > >       
> > >       
> > >     
> >     
> >   
>