ADSM-L

Re: Old problem, new complaints.....

1997-01-27 09:30:58
Subject: Re: Old problem, new complaints.....
From: markus dombrowski <dombrowski.m AT METRONET DOT DE>
Date: Mon, 27 Jan 1997 15:30:58 +0100
LAbbinante wrote:
>
> I know this has been an old standing question/problem with ADSM
> restores.  However,  we recently had a restore of a 60% full 4GB
> filesystem that had 366,000 files in it, and the issue, even though it
> was KNOWN to be a limitation, has come up again.   The file system in
> question was corrupted, and HP, knowing that ADSM was our
> backup/recovery tool, said to reinitialize the filesystem and restore
> from scratch.  After the restore kicked off( which ran at about 28,000
> files/hr.  366,000/28,000=13+hrs), it was brought to my attention that
> files were being missed.  Not files, I found out, but
> directories,....empty directories.   It was decided about three hours
> into the restore to cancel it and recover a fresh unloaded
> (directories/no data) filesystem from other servers.  Needless to say,
> ADSM has been put on the spot for not only for its performance (which
> is due to it's 10mbit connection to MVS), but for the existing problem
> of not restoring empty directories.   Does anyone have an update or
> workaround, on this old, seemingly persistent problem. :-(
> Thanks,,,Larry

We had similar problems with the adsm-netware-client.
Restores with the option "-pi" always skip the directory-entries, the
directories are new created during the restore of the single files, empty
directories are not created. We solved this problem by avoiding the picklist-
option when user-rights and empty directories are involved.
Nevertheless we had another restore problem:
Files which could be seen in the adsm-db weren't restored, when the selection
included multiple files. single restore works fine, but selecting a special 
tree,
always skipped the same files. I never got the technical reason for this bug, 
but
IBM solved this problem with the nw-client level 2.1.06..
I hope this was a help.

markus
<Prev in Thread] Current Thread [Next in Thread>