Re: Help - System Abort 1730 from Subsystem 107

On Wed, 5 Mar 2008 23:23:09 -0500, Dave Bartlet <dave.bartlet@PRODIGY.NET>
wrote:
[color=blue]
>We have a 4way 979 no longer on support which has decided to fail every day
>or 2 with this error. Would anyone know how to determine what the cause of
>this error is? The full msg is:
>System Abort 1730 from Subsystem 107
>System Halt 7, $06C2[/color]

No use asking you to dump and request help from the RC I guess, if this is
no longer on support.
W/o a dump hard to tell but could be you are running out of a table space,
TBLMON or THMGR could tell you more. Popular reason is POSIX files
accumulating too many small extents, frequently that is the diagnostics
logfiles under /var/stm/logs somewhere.

There might be a script on Jazz that helps finding files with many extents,
I would look for files with > 10,000 extents.

Goetz

* To join/leave the list, search archives, change list settings, *
* etc., please visit [url]http://raven.utc.edu/archives/hp3000-l.html[/url] *

too many files in any directory structure (traditional mpe or new)
will bring an mpe system to it's knees.

consider getting 'filecnt' from:
[url]http://jazz.external.hp.com/src/scripts/index.html...and[/url] running it on
a regular basis!

there is a 'numsects' script -- but the no longer needed comment is
worth noting. additionally, if you have mpex, you should be able to
do a listf according to sectors size. - d
--------------------------------
On Fri, Mar 7, 2008 at 7:31 AM, Goetz Neumann <goetz.neumann@hp.com> wrote:[color=blue]
> On Wed, 5 Mar 2008 23:23:09 -0500, Dave Bartlet <dave.bartlet@PRODIGY.NET>
> wrote:
>[color=green]
> >We have a 4way 979 no longer on support which has decided to fail every day
> >or 2 with this error. Would anyone know how to determine what the cause of
> >this error is? The full msg is:
> >System Abort 1730 from Subsystem 107
> >System Halt 7, $06C2[/color]
>
> No use asking you to dump and request help from the RC I guess, if this is
> no longer on support.
> W/o a dump hard to tell but could be you are running out of a table space,
> TBLMON or THMGR could tell you more. Popular reason is POSIX files
> accumulating too many small extents, frequently that is the diagnostics
> logfiles under /var/stm/logs somewhere.
>
> There might be a script on Jazz that helps finding files with many extents,
> I would look for files with > 10,000 extents.
>
> Goetz
>
> * To join/leave the list, search archives, change list settings, *
> * etc., please visit [url]http://raven.utc.edu/archives/hp3000-l.html[/url] *
>[/color]

* To join/leave the list, search archives, change list settings, *
* etc., please visit [url]http://raven.utc.edu/archives/hp3000-l.html[/url] *

Goetz Neumann <goetz.neumann@HP.COM> wrote: There might be a script on Jazz that helps finding files with many extents,
I would look for files with > 10,000 extents.


FILECNT - A script that counts the number of files and directories under any account or directory specified. Wildcards supported. Has been used on large systems with many, many files in a single group. If the script above fails with CIERR 1743 - an IF command nesting problem, please try this version, which is a tarball containing 3 files: a script FILECNT2, a simple NM program that calls the HPCICOMMAND intrinsic, and the source to this NM program. The FILECNT2 version is slower since a process is created for each directory traversal, but should not suffer from an arbitrary nesting limit. Try FILECNT first since it is faster. FILECNT can provide a recursive directory traversal framework that other scripts can utilize.

[url]http://jazz.external.hp.com/src/scripts/FILECNT[/url]

-Craig


* To join/leave the list, search archives, change list settings, *
* etc., please visit [url]http://raven.utc.edu/archives/hp3000-l.html[/url] *