DSXchange

Hi,

I am using Surrogate Key stage in my job to generate a unique sequence number. I am using the STAT file for creating it.

The job worked fine many times previously, but now it started throwing the following error and job getting aborted.

Error:
skg_MatchRecs,0: Unable to lock state file /path/proj1/output/101_CluserFle.txt: No locks available.
skg_MatchRecs,0: The runLocally() of the operator failed.
skg_MatchRecs,0: Output 0 produced 0 records.
main_program: APT_PMsectionLeader(3, node3), player 2 - Unexpected exit status 1.
APT_PMsectionLeader(1, node1), player 4 - Unexpected exit status 1.
main_program: Step execution finished with status = FAILED.

Any one has faced same king of problem before? Any ideas/help to resolve this issue?

--Thanks
--Swathi

Welcome aboard.

What has changed from when it worked and now? Design, total load on the system, configuration, ... anything.

From Job design point of view, we didnt change any thing in the job. But recently I lost all my write/update permissions on LINUX box.

So is it (Permissions) causing the problem?

Probably. When the next block of keys is obtained from the state file information has to be written back to it to record where it "left off".

Ray,

Even after getting the permissions also, job is giving the same error.

Any other thought?

What operating system? Could be a bug at that level...

It is RED HAT LINUX, If I run the job on Single Node config file, it is working fine, but if I execute it on 3 node file, it is giving the error.

Can I use a dataset generated with single node config file in another job running on 3 node config file?

Surrogate stage is working fine now, It is problem with the linux server.

And regarding the Datasets on multinode,
If the dataset is created with multi node and reading with a single node, then it is a probelm. It will throw error like below
has n partitions, but only 1 is accessible from the nodes in the configuration file.

But if the dataset is created with single node and reading from a multi node file, it wont be a problem. Datastage can read the multi node dataset from single node.

Hi SwathiCh

What change you made in Linux level?


Thanks
Man