Parallel jobs on Windows fail with APT_IOPort::readBlkVirt;error

APT_IOPort::readBlkVirt Error Screenshot
APT_IOPort::readBlkVirt Error Screenshot

This a known error for windows systems and applies to DataStage and DataQuality jobs using the any or all the three join type stages (Join, Merge, and Lookup).

Error Message

  • <<Link name>>,0: APT_IOPort::readBlkVirt: read for block header, partition 0, [fd 4], returned -1 with errno 10,054 (Unknown error)

Message ID

  • IIS-DSEE-TFIO-00223

Applies To

  • Windows systems only
  • Parallel Engine Jobs the three join type stages (Join, Merge, and Lookup). It does not apply to Server Engine jobs.
  • Infosphere Information Server (IIS), Datastage and DataQuality 9.1 and higher

The Fix

  • Add the APT_NO_IOCOMM_OPTIMIZATION in project administrator and set to blank or 0. I left it blank so it would not impact other jobs
  • Add the environment variable to the job producing the error and set to 1

What it APT_NO_IOCOMM_OPTIMIZATION Does

  • Sets the use of shared memory as the transport type, rather than using the default sockets transport type.
  • Note that in most cases sockets transport type is faster, so, you likely will not to set this across the project as the default for all job. It is best to apply it as necessary for problematic jobs.

Related References

InfoSphere DataStage and QualityStage, Version 9.1 Job Compatibility

IBM Support, JR54078: PARALLEL JOBS ON WINDOWS FAIL WITH APT_IOPORT::READBLKVIRT; ERROR

IBM Support, Information Server DataStage job fails with unknown error 10,054.

 

*DataStage*DSR_PROJECT (Action=8); check DataStage is set up correctly in project

Error
Error

Basically, the Action=8 error, which I normally see when opening the DataStage Director Client application, means that one or more of the RT_LOG files have become corrupted.  Usually, this problem occurs in relation to disk space issues; although, there can be other causes.

Error Message

Error calling subroutine: *DataStage*DSR_PROJECT (Action=8); check DataStage is set up correctly in project

(Subroutine failed to complete successfully (30107))

The Cleanup approach

The cleanup process really consists of three primary steps:

  • Free disk space
  • Restart application process
  • And, fix corrupted log

Free Disk Space

This can consist of:

  • Cleaning ‘/tmp’ Space
  • Removing any large unnecessary files
  • Enlarging ‘/tmp’ space allocation
  • Adding addition disks space, if necessary

Restart Application Processes

Once you have free the disk space available restarting VM/server is recommended, However, if that is not a realistic option, then at least reboot the Infosphere Datastage engine to ensure the newly freed memory is registering with the applications and to ensure everything is restarted and running.

Fix Corrupted logs

Perhaps, the cleanest way reset the all logs is to perform a ‘Multiple Job Compile’.  Running the jobs will also over write the logs, but is a little more hit and miss, if not all the jobs are not in job streams/batches, which can be run at this time.  The logs can be manually overwritten by compiling the job or performing a reset.  The trick, with manual reset is that you have to know which job to reset, so, this could take a while to get them all. The logs can be manually, dropped and reset, but I recommend that approach only as a last resort.

What does “ERROR: pg_atoi: error in : can’t parse ” mean?

This is one of those vague messages, which takes a little time to run down, but once you understand it, is easily fixed.  Basically, this error happens because you have mismatch and incompatible fields.

Question

What does “ERROR: pg_atoi: error in <ascii>: can’t parse <ascii>” mean?

Answer

The error is the result of an implicit conversion being performed on a character field (varchar/char) while being compared to a numeric field or value.

Vendor Reference Link