Archives of the TeradataForum

Message Posted: Thu, 30 Mar 2000 @ 13:11:24 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

We are running TNT version 03.00.01.04 on an NCR 4400. The data has recently been moved from another Teradata system, and last night I attempted the first backup of a particular database on the new system. When attempting an archive to tape with ARCMAIN, I encountered the first of several errors. The first was when the backup failed, with the message (tablename changed from original):

     03/29/2000 19:02:31 ---  Starting table "TABLE_NAME"
     03/29/2000 19:02:31 ---  Writing table header record
     03/29/2000 19:02:31 ---  Requesting index descriptor from DBS.
     03/29/2000 19:02:32 ---  Writing DBS configuration records
     03/29/2000 19:02:32 ---  Start receiving and writing data block record(s)
     03/29/2000 19:06:49  *** Failure 0306:CLI2: BADPARCEL(306): Invalid parcel received.
     03/29/2000 19:06:49 ---  Disconnecting all sessions.
     03/29/2000 19:06:50  LOGGED OFF  10 SESSIONS
     03/29/2000 19:06:50  ARCMAIN TERMINATED WITH SEVERITY 12

Looking in the Event Viewer, the following appeared just after the ARCMAIN error:

"5148: The Data Block block code or version number is invalid. On Wed Mar 29 19:02:33 2000 on NODE 002-05, VPROC 2, partition 11, task actmain.exe Drive 0 Cylinder 7 Sector 298 for 2560 bytes Datablock failed BLKOK check Blkcode - Checked 108, currently 108, expected 6,7,8" followed by a table listing Proc/Fun, Offset and Line.

At this point, Teradata restarted itself (NT remained running). This morning, we ran scandisk, which stopped after the following error (database/tablename changed from original):

     vproc  2 (0002)  response

     Thu Mar 30, 2000  09:26:40 : DB @ Cylinder 0 7 (0000 0007) Sector 298 (012A) length 5 (0005)
                                : TID :     0  1527  1024 (0000 05F7 0400)  (DATABASE_NAME.TABLE_NAME )
     Thu Mar 30, 2000  09:26:40 : DBD Sector Count Length  2560 (0A00) doesn't match DB Length 53914 (D29A)
     Thu Mar 30, 2000  09:26:40 : SRD TID.uniq doesn't match DB tid.uniq
         0  1527 (0000 05F7)
         1 65280 (0001 FF00)

     Thu Mar 30, 2000  09:26:40 : All other checking of this segment has been skipped
     Thu Mar 30, 2000  09:35:31 : Scandisk found problems

which mentions the same table as the ARCMAIN error mentioned. We then did "check all tables at level one" - during this check, the same 5148 error occurred in the Event Log, and Teradata restarted itself again.

We have tried to drop the table in question (we can restore the data from elsewhere), but it will not be dropped. I tried to select * from the table in question, but this caused another restart of Teradata. Each Event log error always lists the error 5148 in VPROC2, partition 11, Drive 0 Cylinder 7 Sector 298 for 2560 bytes. Those errors which aren't caused by ARCMAIN list "TID: 0 1527 1024 First RID: 8370 28512 0 1 Last Hash: 58909 30212".

Does anyone have any ideas/suggestions? I'm thinking it may be hardware, since it's always on the same sector, but is there any way we can delete the table or move it so that it's not using the bad sector? Any help the list can give would be gratefully received.

TIA,

--
Julie Sorrell

	<Prev	Next>		<<First	<Prev	Next>	Last>>