More Unicode Problems

1999-05-12 08:37:00
Subject: More Unicode Problems
From: Nathan King <nathan.king AT USAA DOT COM>
Date: Wednesday, May 12, 1999 08:37
>I've recently traced an outstanding problem which again appears to be
>related to ADSM's lack of full unicode support:
>We have two clustered NT servers sharing a fibre array of 3 logical drives.
>In order to backup the shared array we have two adsm services installed on
>each server. One service named 'adsm' backups up the local C: and D:
>the other named 'adsmcluster' backs up the clustered drives.
>'adsmcluster' is generically clustered and only started on one server at
>one time. In the event of a failover the 'adsmcluster' service starts on
>other server and backups continue (provided that the backup windows has not
>Last week backups were halting abruptly without error messages, dr watsons
>or anything. The cluster would then restart the adsmcluster service.
>the output appeared as below...
>05/05/1999 23:51:14 ANS1898I ***** Processed   317,500 files *****
>05/05/1999 23:51:14 ANS1898I ***** Processed   318,000 files *****
>05/05/1999 23:51:14 ANS1898I ***** Processed   318,500 files *****
>05/05/1999 23:51:23 ANS1898I ***** Processed   319,000 files *****
>05/05/1999 23:51:34 ANS1898I ***** Processed   319,500 files *****
>05/05/1999 23:51:55 Querying server for next scheduled event.
>05/05/1999 23:51:55 Node Name: RDCLUSTER
>05/05/1999 23:51:55 Session established with server SATADSMPROD1:
>05/05/1999 23:51:55   Server Version 3, Release 1, Level 2.20
>05/05/1999 23:51:55   Server date/time: 05/05/1999 23:51:48  Last access:
>05/05/1999 21:37:26
>I therefore ran an incremental interactively and was surprised to get the
>following error:
>ANS1898I ***** Processed   316,500 files *****
>ANS1898I ***** Processed   317,000 files *****
>ANS1898I ***** Processed   317,500 files *****
>ANS1898I ***** Processed   318,000 files *****
>ANS1898I ***** Processed   318,500 files *****
>ANS1898I ***** Processed   319,000 files *****
>ANS1898I ***** Processed   319,500 files *****
>ANS1898I ***** Processed   320,000 files *****
>ANS1898I ***** Processed   320,500 files *****
>Normal File-->       110,608,384
>Assertion failed: CB_HDRP((char *) blkP - FB_SIZE(blkP))->prevFree == 0,
>file d
>mem.c, line 1512
>abnormal program termination
>After much searching I managed to ascertain that infact the adsm client was
>abending when attempting to backup files that had characters such as | in
>the filename. NT guru's may know that the character | is deemed to be an
>illegal character in an NTFS filename. In actual fact what this filename
>contains is a character mapped into the private use area range of unicode.
>Windows NT chooses to represent private use characters within Explorer
>characters such as | . Since ADSM is not fully unicode compliant it can't
>backup files that contain characters mapped into the private use area
>Under normal circumstances ADSM simply ignores these files and generates
>error. ANS1256E Cannot make file/directory. This abnormal program
>termination is a new one on me though. Has anyone seen anything similar???
>Does anyone know of a fix other than deleting the offending files or
>off unicode support completely?
>Nathan King
>Open Systems Storage Management
>*  (210) 913-5755
>* Pager: 753-5755
>* E-mail: nathan.king AT usaa DOT com
>The opinions herein are mine and almost certainly do not reflect those of
<Prev in Thread] Current Thread [Next in Thread>