DSXchange

hi,

The job has two udb DB2 tables and they are inner joined using a join stage. The input links to the join are hashed and sorted on the join key. This job was running without any issues for the past one month. But now I got the following error:

JoinGst.LkRefDb2udbGst_Sort,0: Failure during execution of operator logic.
JoinGst.LkRefDb2udbGst_Sort,0: Input 0 consumed 0 records.
JoinGst.LkRefDb2udbGst_Sort,0: Output 0 produced 0 records.
JoinGst.LkRefDb2udbGst_Sort,0: Fatal Error: Tsort merger aborting: stat() of sorted run file failed
JoinGst.LkRefDb2udbGst_Sort,0: Failure during execution of operator logic.
JoinGst.LkRefDb2udbGst_Sort,0: Input 0 consumed 5146321 records.
JoinGst.LkRefDb2udbGst_Sort,0: Output 0 produced 0 records.
JoinGst.LkRefDb2udbGst_Sort,0: Fatal Error: Pipe read failed: short read
node_prod01_node1: Player 2 terminated unexpectedly.

I am using a 16 node configuration file.

what could be the possible reason for the error "Fatal Error: Tsort merger aborting: stat() of sorted run file failed"?
Is this because the operator could read data using the DB2 UDB stage but could not write the sorted data on to a temporary file or a scratch disk on unix.
Is the Short Read error due to the cascading effect or is it an independent error which caused the player 2 to terminate abnormally?

I re ran the job but it failed with the following error:

main_program: File archive: Trouble creating file "/datastage/tmp/APTcs2523218f7dea56d"
main_program: Fatal Error: Null archive.


When I am trying to view data from the tables it shows the Null Archive error. This Null Archive error is shown on almost all the tables.

I presume it might be an environment issue.
Kindly provide proper insights.

Hi Madhavan,

Pls check whether the scratch disk mentioned in your configuration file has enough space.

If possible increase the scratch disk space, run the job and report what happens.

--Rich