As you can see, the total exec time
for any past runs is derived from the RemoteWallclockTime minus the CumulativeSuspensionTime
as described in the manual's description of the former attribute, and then
this is added to the total time accumulated so far in the current run,
if the job is actually running.It's necessary to use time from both the
current and any prior runs because the RemoteSysCpu and RemoteUserCpu attributes
do too.
However, this falls apart if the RemoteWallClockTime
is updated before the job finishes running. From my testing so far, it
doesn't appear to be, but if anyone knows for sure, or is familiar enough
with the code to find it, I'd appreciate a tip.
Michael V. Pelletier
IT Program Execution
Principal Engineer
978.858.9681 (5-9681)
339.293.9149 cell
michael.v.pelletier@xxxxxxxxxxxx