LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: HB declares the local machine dead and kills LD

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: HB declares the local machine dead and kills LD
From: Horms <horms@xxxxxxxxxxxx>
Date: Mon, 17 Nov 2003 10:56:06 +0900
On Sun, Nov 16, 2003 at 07:19:46AM -0500, John Barrett wrote:
> 
> ----- Original Message ----- 
> From: "Lars Marowsky-Bree" <lmb@xxxxxxx>
> To: "LinuxVirtualServer.org users mailing list."
> <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
> Sent: Saturday, November 15, 2003 4:13 PM
> Subject: Re: HB declares the local machine dead and kills LD
> 
> 
> > On 2003-11-15T11:54:23,
> >    John Barrett <jbarrett@xxxxxx> said:
> >
> > > I'm using 1 second heartbeats with a 5 second dead time, but still from
> time
> > > to time I get THIS:
> > > (note: the machine declared dead is the local machine !!!)
> >
> > There is a kernel bug in some kernel releases which prevents heartbeat
> > from being scheduled often enough.
> >
> > Which kernel are you running?
> >
> > The workaround is obvious, use a longer (maybe 10s) deadtime.
> >
> >
> > Sincerely,
> >     Lars Marowsky-Brée <lmb@xxxxxxx>
> >
> 
> latest UultraMonkey kernel for RH9 -- I've got dead time cranked to 20
> seconds, but thought someone might like to chase the bug :)

Which incidently should be the latest RH9 kernel from Red Hat
recompiled after adding in LVS and the hidden interface patch.
I made no other changes to the kernel packages.

-- 
Horms
<Prev in Thread] Current Thread [Next in Thread>