candidate host pending reason

It's about pending reason hostname.

Current pending reason as below. These 3 hosts are not in host group of this queue, but they are in candidate host list.

PENDING REASONS:
Candidate host pending reasons (18 of 7574 hosts):
Job slot limit reached: pivy2365, hsandy1793, hsandy1784, hsandy1780, pivy03
09, pivy1960, pivy0706, pivy0705, pivy0704, pivy0703, pivy
0144, hsandy1803, pivy0142, hsandy1855, hsandy1846;
Load threshold reached (Resource: rpcport): pivy0134, pivy1983, pivy0139;

So the customer expects the following pending reason.

PENDING REASONS:
Candidate host pending reasons (18 of 7574 hosts):
Job slot limit reached: pivy2365, hsandy1793, hsandy1784, hsandy1780, pivy03
09, pivy1960, pivy0706, pivy0705, pivy0704, pivy0703, pivy
0144, hsandy1803, pivy0142, hsandy1855, hsandy1846;


We think hosts not included in the queue should not be listed.

But I have learned the following:

Based on current LSF design, "Load threshold reached" is marked as candidate host pending reason.
But when these 3 hosts are unavailable, their pending reason changes to "Load information unavailable" that will be marked as non-candidate pending reason. LSF consider a pending reason as candidate or non-candidate based on pending reason content, instead of host name.
In the final analysis, customer hopes to filter all pending reasons (no matter candidate or non-candidate) based on host name, instead of pending reason category. It will impact not only displaying, but also scheduling performance.
Originally we also thought it was a defect. But after developer's further investigation and discussion with offering manager (original product manager), filtering host name for a queue for ALL pending reasons is a new requirement. We cannot fix it as a bug because it will change LSF design and default behavior. Our development need to consider globle customers.

We understand that
LSF consider a pending reason as candidate or non-candidate based on pending reason content, instead of host name.

But from the user's point of view,
They don't need to see these hosts if host is not in specified host group.

So we hopes to filter all pending reasons (no matter candidate or non-candidate) based on host name, instead of pending reason category.
Admin just want to see only hosts in related queue.
It seems reasonable.