Hi Steve,
thanks for the answer. See a more detailed description below.
On Mon, 21 Jan 2002, Steve Best wrote:
> > Jan 17 03:08:50 machine kernel: DT_GETPAGE: dtree page corrupt
> > Jan 17 03:08:50 machine last message repeated 4 times
> > Jan 17 03:08:50 machine kernel: jfs_lookup: dtSearch returned 5
> >
> > And when I just did a manual updatedb I got hundreds more, together
> > with messages that some files could not be found, e.g.:
> >
> > Jan 17 10:19:34 machine kernel: DT_GETPAGE: dtree page corrupt
> > Jan 17 10:19:34 machine kernel: jfs_lookup: dtSearch returned 5
> > Jan 17 10:19:34 machine kernel: DT_GETPAGE: dtree page corrupt
> > Jan 17 10:19:34 machine kernel: jfs_lookup: dtSearch returned 5
>
> > I am running a Dual-PPro200 system with Kernel-2.4.14+loop-patch and
> > jfs-1.0.12 on SuSE-7.1.
>
> I see that you have a loop-patch on your kernel, is JFS being used
> as a loop device here?
No, I do compile the loop-device but do not use it at the moment.
> Could you tell us more about your applications that are running on this
> system, so we can try to re-create this problem here. This type of problem
> is easier to fix if we can debug it here.
I am running pine on the system all the time (probably not the
problem :-) ).
The messages at 03:08:50 were related to TSM-4.1.1 doing the backup
of the machine (starting exactly, which probably does some kind of
find. For that purpose the JFS-Partitions (mounted on /users and
/users2) are both set up as VIRTUALMOUNTPOINT so that it recognizes
them. And I just found an error in its log, too:
Executing scheduled command now.
01/17/2002 03:06:32 --- SCHEDULEREC OBJECT BEGIN DAILY_INCR 01/17/2002 03:00:00
01/17/2002 03:08:50 --- SCHEDULEREC STATUS BEGIN
01/17/2002 03:08:50 Total number [...]
[...]
01/17/2002 03:08:50 --- SCHEDULEREC STATUS END
01/17/2002 03:08:50 ANS4023E Error processing '/users': file input/output error
01/17/2002 03:08:50 --- SCHEDULEREC OBJECT END DAILY_INCR 01/17/2002 03:00:00
Otherwise I was was working remotely via NFSv3 on some files of
sizes around 4 MB (astromonical data).
And the final hang occured while I was doing an updatedb with
localpaths /home and /home2, which are autofs "mounted" as links to
/users and /users2. It also complained about some files not being
accesible.
For completeness: the / is still ext2.
> If you have a test case that you can send us to create this one please do.
Unfortunately, I have seen this problem before, but I cannot really
recreate it intentionally. The days since then I have not seen these
messages although the same apps were running. But that I see them in
correlation with some kind of find makes me suspicous...
Peter.
_______________________________________________
Jfs-discussion mailing list
[EMAIL PROTECTED]
http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jfs-discussion
