HP3000-L Archives

January 2001, Week 3

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
"VANCE,JEFF (HP-Cupertino,ex1)" <[log in to unmask]>
Reply To:
VANCE,JEFF (HP-Cupertino,ex1)
Date:
Fri, 19 Jan 2001 22:38:37 -0800
Content-Type:
text/plain
Parts/Attachments:
text/plain (38 lines)
> In the middle of a "LISTF GETYR@,2" command she got the
> following display:
>   Cannot access file "/ACCT/GROUP/GETYR3FN" due to lock
>   contention  (CIWARN 9169)
>
> I remember someone reporting something similar when dealing
> with one of the 'newer' 'listf' options, but nothing with the
> old mode 2.

LISTFILE (and LISTF) were partially re-written to support the
"access" (,8 and ,9) feature.  One of the goals for LISTF,access
was to not have the LISTF hang if there was a real problem with
the file being listed.  The original code did a special open of
each file before listing it.  This open unconditionally locked a
few file system semaphores.  A file system bug (or some very unusual
condition) could cause the open to wait forever and the LISTF would
hang.

The new code doesn't really open the file, but still needs to obtain
the same semaphores.  However, it locks one of the semaphores with a
short time out value (10 millisec), so if that semaphore is busy, LISTF
won't get the lock that time.  The code loops 400 times trying to get
the lock with the 10 millisec pause after each lock attempt.

Typically, you won't see CIWARN 9169.  There was a patch or two to
increase both the pause time and the number of iterations.  The
numbers I used above are for 6.5 and later.  The new LISTF locking
method affects all formats except 0 and 6.

Probably an SR should be submitted to the lab, since the goal of the
locking implementation was (and is) to never see CIWARN 9169 unless
you have a file hang situation.  I've cc'd Craig Fairchild (File System
Guru), who helped with this design and, in case, I have over simplified
the explanation.

Hope this helps,
 Jeff Vance, CSY

ATOM RSS1 RSS2