On Mon, 8 Sep 1997, Booher, Joelle M wrote:
> Good Monday morning!
>
> I am running ADSM Server for NT, PTF3. This weekend one of our clients
> got the following error:
>
> 09/06/1997 03:00:56 Expiring--> 426,975
> E:\data\Mpro\CHUCK.$$D Sent
> 09/06/1997 03:00:56 Expiring--> 17,348
> E:\data\Mpro\CHUCK.$$X Sent
> 09/06/1997 03:00:58 ANS4102I ***** Processed 227,500 files *****
> 09/06/1997 03:01:04 ANS4102I ***** Processed 228,000 files *****
> 09/06/1997 03:01:06 ANS4102I ***** Processed 228,500 files *****
> 09/06/1997 03:01:10 ANS4102I ***** Processed 229,000 files *****
> 09/06/1997 03:01:28 ANS4253E File input/output error
> 09/06/1997 03:01:28 Incremental backup of drive 'F:' with label
> 'TYPHOONF'
> 09/06/1997 03:01:38 Directory--> 0 F:\INSTALLR Sent
>
> As you can see, it was backing up drive E: at the time that the error
> occurred, then went right to drive F:. Unfortunately there are about
> 10,000 more files on drive E: that it never attempted to backup. Upon
> investigation, it appears that there is a corrupt file in one of the
> subdirectories on drive E: which cannot be deleted. Microsoft has
> instructed the admins of that NT server that they will need to run a
> chkdsk -f, and reboot the server before the file gets cleaned up. The
> admins have informed me that they can not bring the server down until
> next weekend. AND, to complicate matters, they do not feel comfortable
> running that chkdsk unless they have good backups - which they are not
> getting because of the I/O failure. I ran a selective of all of those
> directories so that they would have backups of Friday's updates, but I
> can't continue to do that all week because it's causing good versions to
> be pushed out, and it's flooding my already 14Gb database!
>
> Here's my question (finally!): How do I get ADSM to totally ignore the
> corrupt file so that the rest of the files on that drive will be backed
> up? It appears that even if I exclude the file, ADSM tries to read it
> or something because I still get the error/failure.
>
> Any suggestions?
>
> Thanks in advance for your help....
> Joelle Booher
> Amp Incorporated
> (717)810-2816
> jmbooher AT amp DOT com
>
Hi,
given the file causing the problems resides in a subdirectory and the
rest not getting backup'ed resides further down or in different
subdirectories you could either ( both are entries in "dsm.sys" )
1. introduce VIRTUALMOUNTpoint-s which point to directories NOT including
the problem case ( unfortunately only applicable to UNIX-systems,
accordingly to the "Installing the Clients" book ).
BTW: does the DOMAIN-option ( under NT ) allow path-specifiers ?
2. just move the problem-file to a different point in the tree within a
PRESCHEDULE and move it back to the old position within a POSTSCHEDULE
( the following example tackled a severe performance problem )
*
*pwd 04feb96: Deal with tens of thousands of links in .../tubs/tmp -directory
*
PRESCHED "mv /dsk/rzis2.scsi5/harvest/gatherers/tubs/tmp /dsk/rzis2.scsi5"
POSTSCHED "mv /dsk/rzis2.scsi5/tmp /dsk/rzis2.scsi5/harvest/gatherers/tubs"
The above is working on UNIX-systems.
--
MfG / Ciao - - - - - - - - - - - - - - - - - - - - - - - - - - - -
MfG / Ciao - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Peter Dümpert Email: p.duempert AT tu-bs DOT
de
Rechenzentrum der Technischen Universität Fax : ++49/531/391-5549
D 38092 Braunschweig Tel : ++49/531/391-5535
|