Conference turris::digital_unix

9866.0. "rpc.pcnfsd dies ..." by TKTVFS::ASARA (Investigate well, then open the OUTAGE.) Mon May 19 1997 05:12

Hi,

Our customer is experiencing the trouble of rpc.pcnfsd.
We know that various rpc.pcnfsd problems are written to this conferense.
(ex: 5855, 8808, 9420, 9609) Those are similar, but aren't exactly same as
our case.

The problem is rpc.pcnfsd frequently dies with and without core.
When it leaves core, we can get the following stack trace.

	0 siad_ses_authent() ...
	1 __sia_ses_authent() ...

Customer uses Digital UNIX 3.2D-1 on their AlphaServer. We installed
OSF360-350223 that contained rpc.pcnfsd but it didn't fix the problem.

We can't get any abnormal messages from syslog but we can find curious
message in sialog. We believe that the timestamp of sialog is exactly same as
the timestamp of core file. (We can't pinpoint the timing of coreless problem.)

	SIA: EVENT Thu Apr 24 08:28:57 1997
	Successful session authentication for ABC on (null)
	SIA: ERROR Thu Apr 24 08:28:57 1997
	Failure to establish session

This messages is really curious. Because 'Successful' and 'Failure' were
logged on the same time.

From above stack trace, we believe that rpc.pcnfsd dies due to some SIA
inconsistency and above sialog relates with the problem. We are asking
customer who ABC is. What does ABC do ?

On the other hand, we found QAR.46407 that relates with rpc.pcnfsd.
Another rpc.pcnfsd was provided as the solution. We looked for the location of
this solution. But it seems already removed.

From the answer of this QAR, we got the revision of updated files in rpc.pcnfsd.

    ./usr/sbin/pcnfsd/pcnfsd_misc.c  Rev. 1.1.17.2     (backport of Rev. 1.1.12.2)

   ./usr/sbin/pcnfsd/pcnfsd_svc.c  Rev. 1.1.7.2       (backport of Rev. 1.1.6.2)
   ./usr/sbin/pcnfsd/pcnfsd_svc.c  Rev. 1.1.7.2

   ./usr/sbin/pcnfsd/pcnfsd_v1.c  Rev. 1.1.13.2 
   ./usr/sbin/pcnfsd/pcnfsd_v2.c  Rev. 1.1.13.2 

And we got the following revsions from OSF360-350223 rpc.pcnfsd.

# what rpc.pcnfsd
rpc.pcnfsd:
        $RCSfile: crt0.s,v $ $Revision: 1.1.24.3 $ (DEC) $Date: 1995/03/14 16:57:40 $
        $RCSfile: pcnfsd_svc.c,v $ $Revision: 1.1.2.3 $ (DEC) $Date: 1992/09/29 14:19:47 $
        $RCSfile: pcnfsd_xdr.c,v $ $Revision: 1.1.2.3 $ (DEC) $Date: 1992/09/29 14:20:21 $
        $RCSfile: pcnfsd_v1.c,v $ $Revision: 1.1.10.2 $ (DEC) $Date: 1996/04/25 00:28:33 $
        $RCSfile: pcnfsd_v2.c,v $ $Revision: 1.1.10.2 $ (DEC) $Date: 1996/04/25 00:28:56 $
        pcnfsd_v2.c     1.6 - rpc.pcnfsd V2.0 (c) 1991 Sun Technology Enterprises, Inc.
        $RCSfile: pcnfsd_misc.c,v $ $Revision: 1.1.23.2 $ (DEC) $Date: 1996/06/10 17:08:21 $
        $RCSfile: pcnfsd_cache.c,v $ $Revision: 1.1.2.3 $ (DEC) $Date: 1992/09/29 14:19:16 $
        $RCSfile: pcnfsd_print.c,v $ $Revision: 1.1.22.2 $ (DEC) $Date: 1996/06/10 17:08:23 $


From this comparison, it seems to us that rpc.pcnfsd of QAR solution is recent
than OSF360-350223.

At this moment, we'd like to know the following things.
Could anyone please advise to us ?


	1. We assume that QAR solution is recent than OSF360-350223.
	   Is this correct ?

	2. If so, where can we get it from ?

	3. If we should open IPMT case, what kind of information we should
	   gather especially client PC ?
	   (We've already gathered information of server machine by using
	    the checkit script.)

Thank you and best regards,
Satoshi Asara