SUMMARY: HELP: fcntl Diagnostic

From: Anthony Yen (tyen@mundo.eco.utexas.edu)
Date: Fri Sep 03 1993 - 19:03:55 CDT


Original Message:

> I have a SPARCsystem 670MP running SunOS 4.1.3 and 64Mb RAM. It is
> repeatedly sending the following diagnostic to the console:
>
> fcntl: Invalid argument
> rpc.lockd: Unable to set a lock.
>
> Killing rpc.lockd and restarting it only pauses the output for awhile,
> and then it starts anew again. How can I track down which offending
> process is causing this?
>
>I can restart the machine of course, but then that doesn't guarantee
>me that the problem is permamently solved. I'll summarize back to the
>list, of course.

Respondents:

        steve@gec-epl.co.uk (Steve_Kilbane)
        glenn@uniq.com.au (Glenn Satchell - Uniq Professional Services)
        kevin@uniq.com.au (Kevin Sheehan {Consulting Poster Child})
        djiracek@jupiter.fnbc.com (Dan Jiracek)
        Victor.Fieldhouse@mail.esrin.esa.it (Victor Fieldhouse)
        poffen@sj.ate.slb.com (Russ Poffenberger)
        

Summary:

Almost everyone said to get patch 100075-xx. The patch level of the
patch varied from a low of -06 to a high of -13. I've found out that
-13 is the latest, so those who thought it was less than this might
want to check out their favorite patch archive sites again.

One respondent said:

> I don't know if this will help but when we had some rpc.lockd errors
> we had to remove the contents of /etc/sm and /etc/sm.bak

steve@gec-epl.co.uk (Steve_Kilbane) gave an explanation of the problem:

        the fault is in the lockd, and is a known bug. the bug is that
        when you have two machines with an nfs mount between them, and
        a remotely-locked file gets deleted by a lock process, lockd
        replies to the remote process that the lock could not be
        obtained, rather than the file is not present.

        the fix is to get a new lockd from sun.

        the weird part is that this problem was fixed in 4.1.2's lockd
        - which means you must have 4.1.1 clients screwing things up.

And indeed, the short term fix I applied was to kick my 4.1.1 clients
up to 4.1.3; I'm still integrating this new patch into my patch kit.

Many thanks to everyone who responded!

Anthony Yen - SPARC Sysadmin - Department of Economics - UT/Austin
ECB 3.138, MS#63100 - 471-3211 ext 170 - tyen@mundo.eco.utexas.edu



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:08:09 CDT