"Visu" <Vi*******@gmail.com> a écrit dans le message de
news:11*********************@g44g2000cwa.googlegro ups.com...
Hi all,
DB2 - Ver 7.2 EEE on AIX
This Sytem temporary tablespace in question is created on all the 5
nodegroups, and the containers are in 5 different filesystems. For some
reason only one container is being used and eventually the filesystem
is getting FULL and the transcation fails.
The SQL being executed is a select statement with about five joins and
a group by , and tables involved are relatively big.
The filesystem holding each containers have 37GB freespace. And ulimit
for the user is unlimited.
Any hint on why this happens will be a great help!
Thanks
Hi, you should look at the access plan to see if at one step all goes to a
mono-partition nodegroup thru a tablequeue, to join with a table that is in
this mono-partition nodegroup.
One way to lift this is to create a REPLICATED table of this table in a
tablespace that is multi-partition to allow DB2 continue inter-partition
parallelism. Don't forget to runstats on this table, and you can also create
indexes (non unique) on this type of table.
HTH,
Jean-Marc