"Bad container path" during Backup?

[Norbert Munkel]

Hi,

System: 8.1.5 UDB on SuSE Linux

I created a new SMS tablespace with following ddl:

CREATE REGULAR TABLESPACE ASNCTL IN DATABASE PARTITION GROUP
IBMDEFAULTGROUP PAGESIZE 4096 MANAGED BY SYSTEM
USING ('/data/db2/db2inst1/ASNCTL')
EXTENTSIZE 32
PREFETCHSIZE 32
BUFFERPOOL IBMDEFAULTBP
OVERHEAD 12.670000
TRANSFERRATE 0.180000
DROPPED TABLE RECOVERY ON;

Tablespace-ID is 10.

Well, it works fine and everything. No errors at all but during full
database backups.
I can do a online tablespace backup for ASNCTL which works fine.
At full database backups I get a SQL0298N. As this is the only
database at that instance and the path is read-write accessible, the
documentation of 0298 is reduced to

"This message will also be returned if any other unexpected error occurred
which prevents DB2 from accessing the container."

I moved the whole tablespace to another location for testing which did not
help.

db2diag.log looks like this:

---snip----

2004-12-01-01.13.44.772003 Instance:db2ins t1 Node:000
PID:19572(db2ag ent (PRODDB)) TID:1024
Appid:*LOCAL.db 2inst1.05068022 0301
buffer pool services sqlbTestSetData poolState Probe:633 Database:PRODDB

poolID = 10, oldState = 0, newState = 800, currentState = 0

2004-12-01-01.13.44.790401 Instance:db2ins t1 Node:000
PID:19572(db2ag ent (PRODDB)) TID:1024
Appid:*LOCAL.db 2inst1.05068022 0301
buffer pool services sqlbTestSetData poolState Probe:778 Database:PRODDB

DIA0001E An internal error occurred. Report the following error code :
"ZRC=0x8702007D ".

2004-12-01-01.13.44.791493 Instance:db2ins t1 Node:000
PID:19572(db2ag ent (PRODDB)) TID:1024
Appid:*LOCAL.db 2inst1.05068022 0301
database utilities sqlubHandleBack upInProgressAnd RebalOnError Probe:1
~ Database:PRODDB

Cannot turn BACKUP_IN_PROGR ESS off for pool:

0x64F5ACF8 : 4153 4E43 544C ASNCTL

2004-12-01-01.13.44.819137 Instance:db2ins t1 Node:000
PID:19572(db2ag ent (PRODDB)) TID:1024
Appid:*LOCAL.db 2inst1.05068022 0301
database utilities sqlubHandleBack upInProgressAnd RebalOnError Probe:1
~ Database:PRODDB

DiagData
0x64B58870 : 0x0000007D87020 07D }...}...

2004-12-01-01.13.44.826046 Instance:db2ins t1 Node:000
PID:19572(db2ag ent (PRODDB)) TID:1024
Appid:*LOCAL.db 2inst1.05068022 0301
database utilities sqlubHandleBack upInProgressAnd RebalOnError Probe:1
~ Database:PRODDB

0x64B58870 : 0x0000007D87020 07D }...}...

2004-12-01-01.13.47.016963 Instance:db2ins t1 Node:000
PID:19572(db2ag ent (PRODDB)) TID:1024
Appid:*LOCAL.db 2inst1.05068022 0301
database utilities sqlubcka Probe:140 Database:PRODDB

Backup Terminated.

---snap---

I understand that the backup tool tries (successfully?) to switch the
tablspaces state from 0 (normal) to 800 (backup_in_prog ress) and back
again. Seems to fail somewhere in between.

I wonder If some of you has seen that behaviour before and could give me a
hint how to handle it. The db is holding about 400 GB of data and it
always crashes at the end of the backup process.

regards,

Norbert

[Norbert Munkel]

Hi,

Norbert Munkel wrote:

Well, it works fine and everything. No errors at all but during full
database backups.

IBM support told me, it happened to at least one other customer as well.
The first scheduled backup after the tablespace creation failed with
"Device Full". According to the supporter, the tablespaces state was
changed to "backup in progress" although "list tablespaces" still
reported it as normal.

I finally got the backup working again by moving ASNCTL to different
location and changing the tablespace type to DMS.

The recommended solution by IBM support required database stop, database
start, activate datbase, backup tablespace, backup database.

Thanks anyway. Maybee it can be helpfull for some poor guy in the future
if he can google this. ;-)

regards,

Norbert