When the last psom job fails, the pipeline execution sometimes return success du to a racing condition related to the garbage collector.
The best solution it seems it to create a PIPE.exit_code when the PIPE.lock is deleted. (but what should be the ret status when the PIPE.lock is manually removed? "2", maybe "128" ?
To flush the nfs cache in at directory, to most straightforward method is to do "ls /dir/to/flush/nfs"
When the last psom job fails, the pipeline execution sometimes return success du to a racing condition related to the garbage collector.
The best solution it seems it to create a PIPE.exit_code when the PIPE.lock is deleted. (but what should be the ret status when the PIPE.lock is manually removed? "2", maybe "128" ?
To flush the nfs cache in at directory, to most straightforward method is to do "ls /dir/to/flush/nfs"