Job array subjobs are not being held after failing/requeueing 20 times.

[openpbs]

Job array subjobs are not being held after failing/requeueing 20 times.

[dilip-krishnan]

The question that comes here is, should we just held the failing subjob or the whole array job. Personally
I feel we should maintain the fail count in the parent job and at any point if the cumulative failure of subjobs reaches to 20 we should held the whole array job instead of putting a hold on the failing subjob.

This actually reminds me of a forum discussion on how the array job is percieved in PBS. AFAIK we consider array jobs as a part of single normal job, instead of considering each subjob as a normal job. Here is the link to the discussion http://community.pbspro.org/t/understanding-array-jobs/212/2 .