For more accurate pend time

Product: IBM Spectrum Computing Analytics
Version: 9.1.4
Actor: SKhynix, Korea HQ

Customer want to know how many server's slot was not used and application license was pended and what user could not run own jobs in how long time.

To find some information, customer are using some information between LSF and PA.
It is the output of bhist command in LSF side and it is RPT_JOBMART_RAW and RPT_JOB_PENDING_STATUS table.

To check pending time and pending reason for each job, we always compare pending time between LSF and PA information.
As you know, for pending time, the output of bhist command is always same with the value in RPT_JOBMART_RAW table.
(LSF's job pending time = PA pending time in jobmart table)

But, the problem is, sometimes RPT_JOBMART_RAW's pending time is not matched with RPT_JOB_PENDING_STATUS's value.
Sometimes, the value in RPT_JOBMART_RAW table is grater than RPT_JOB_PENDING_STAUTS and sometimes, the value in RPT_JOB_PENDING_STAUTS is greater than RPT_JOBMART_RAW.

So customer thinks this is not accurate each other and the value in RPT_JOB_PENDING_STAUTS is not correct, because each pending value between LSF and RPT_JOBMART_RAW is matched.

So, we tried to find some reason and discussed with IBM support team about this. And then we know there are some external and interval reason, as like NTP problem or the time(date) is different between LSF master host and PA node.

So, as like this situation, to make them more correct, I will suggest the new flow.