[Linux-HA] issue with management of heartbeat.pid file

[Andrew Beekhof]

On 5/24/07, Brian Reichert <reichert at numachi.com> wrote:
> On Thu, May 24, 2007 at 12:21:13AM -0700, Kevin Jamieson wrote:
> > Brian Reichert wrote:
> >
> > >What I tracked down was that if the box powered down too quickly
> > >for heartbeat to clean up, a PID file was left in place:
> > >
> > >...
> > >
> > >But, there's no check to assure the recorded PID is not stale.
> > >
> > >Have others seen this?  This code seems to be in 2.0.8 as well...
> >
> > FYI, I filed a bug on this a few months back:
> >
> > http://old.linux-foundation.org/developer_bugzilla/show_bug.cgi?id=1454
> >
> > There's a patch attached to that bug that worked for us (although the
> > patch is Linux-specific, so may not be ideal).
>
> If heartbeat is running as root at the time,  ( kill(pid,0) != -1
> ) implies that pid is currently in use.
>
> I don't see what the checks in /proc gets you at this point; how
> is that different than
>
>    ( kill(pid,0) != -1 ) && pid !=getpid() )
>
> I'm not used to using /proc for this stuff

it tells you if the pid has been reused by another, unrelated, process.
its just an extra check available to linux users.

i committed a modified patch earlier today