http://defect.opensolaris.org/bz/show_bug.cgi?id=12567
amaguire <alan.maguire at sun.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |alan.maguire at sun.com
--- Comment #3 from amaguire <alan.maguire at sun.com> 2009-11-10 16:28:09 UTC
---
(In reply to comment #2)
> I have found that this bug exists even after the fix bug 12462 nameservices
> need to be online before we restart autofs. This bug does not occur with the
> Automatic location, but occurs with the User location. My assumption here is
> that the NIS configuration is the User location is causing the problem.
>
> Further investigation has led me to:
> - there are two instances of ypbind running - one that existed when everthing
> was working fine. A new ypbind process is added.
>
> - SMF reports nis/client as online, but continues to start new ypbind
> instances
> when the most recently started ypbind process dies (evident from the
> increasing
> process ids of ypbind). ptree shows this process as the child of the original
> ypbind process
>
> - restart the nis/client service is a workaround that brings up the interfaces
> and there is only one ypbind process.
can you do a pstack on the two ypbind processes? If I remember, the :kill stop
method will send a SIGTERM, but the following comment in the ypbind source
makes me wonder if there's an issue with signal handling in ypbind:
604 /*
605 * This routine is called when we are killed (by most signals).
606 * It first tries to unregister with the portmapper. Then it
607 * resets the signal handler to the default so that if we get
608 * the same signal, we will just go away. We clean up our
609 * children by doing a hold in SIGTERM and then killing the
610 * process group (-getpid()) with SIGTERM. Finally, we redeliver
611 * the signal to ourselves (the handler was reset to the default)
612 * so that we will do the normal handling (e.g., coredump).
613 * If we can't kill ourselves, we get drastic and just exit
614 * after sleeping for a couple of seconds.
615 *
616 * This code was taken from the SunOS version of ypbind.
617 */
--
Configure bugmail: http://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
