DSXchange

DataStage has been partitioned to see 4 cpu's (2 hyperthreaded). DataStage sees all 32 cpu's. Yep the Unisys box is not bad, one can partition it to the point of having a different OS running each partition. It's like splitting the box into many servers. We have not done that on this box. We have a ...

Thanks Roy ... encouraging it is not !!

I guess I'm going to have to beg Ascential for some freebies

Hi Roy According to the guys who set up the partition, it is independent, hence my dilema. Also, call has been logged with Ascential by no joy there either. We use job sequencers and the warning status from the license warning is causing problems because of the warning triggers we have set. I will s...

We have a 32 cpu machine that has been partitioned. DataStage runs under a 4 cpu partition because we have a 4 cpu license. However, when the jobs are run we get warnings in the job logs about violation of the number of cpu's. It seems DataStage does a 'system' scan instead of a partition scan. I ha...

Thanks Guys...

Will try the cleanup as per Ray's instructions.

Hi Guys Thanks for all the replies. What seemed to be the problem is the name of the distributed file i.e. FT_AKANAMESHASH When I changed the name to FT_AKANAMES_HASH and used INTERNAL "IF FIELD(@ID,@TM,4) < 4000000 THEN 1 ELSE etc it worked. I can view data in the distributed file, so all is a...

Me again. I tried all the options, still get the error about the partitioning algorithm for the distributed file not matching when I run the DEFINE.DF command. I also tried: INTERNAL "IF FIELD(@ID,@TM,4) < 4000000 THEN 1 ELSE etc -- no joy What is the correct procedure for creating a distribute...

1. The job creating the hash files is a multi-instance job. 2. Each instance of the job is passed different parameters for a sql select statement (where akakey >= startkeyvalue and akakey < endkeyvalue). 3. The hash file contains 4 key columns, akakey being the last of the 4. 4. Each key column is d...

No sure what you mean Wolfgang. I understand that each hashfile's metadata has to be the same. Here's my dilema...if I use @inrownum to send 1000 rows to hash file 1, the next 1000 rows to hash file 2 etc, how do I specify this in the DEFINE statement when I'm creating the distributed hash file. Can...

Hi Guys I have created a job that creates 10 hash files. Each hash file will be incorporated into a distributed hash file. The average number of rows in each hash file is 6 million. The hash file has 4 columns that make up the key, the constraint in the job is based on the value in only one of the h...

Thanks very much for the replies...I'm a bit embarassed (The option was there all the time. I guess we all need to look a little harder at the screen sometimes...)

I have a job that is marked as a muti-instance job (in the job properties). The job executes without any problems. However, when I try to invoke another instance of the job, it does not allow it (job is already running message). In the Director, the run button is greyed out. :x Do I have to set this...

Hi Vincent I'm not sure I understand what you mean when you say the splitter will only run one cpu. From my understanding, each link out of the splitter will force a new 'thread' which, in turn, will mean the cpu's will be better utilised :? . If this is not the case, then from a performance enhanci...

Thanks guys. I fiddled with the key columns and now it works.

Interesting though, I tried the job using the splitter and then created another job using the IPC stage without the splitter. The IPC stage beats the splitter every time.

I am updating some key columns. However, the "Where" statement refers to the primary key only.

Does DB2 not allow updating of non primary key columns