Issue:
The statistics for mstats are labeled with cryptic names, and are not descriptive
Affected Version:
All currently supported releases
Problem:
Running the mstats --xml command provides a lot of statistics, but no descriptions.
Solution:
The following list of fields should provide this information
Field ID | Field Description |
ABP | Job weighted average bypass count, Bypass / JC |
APS | active job proc-seconds remaining |
AQT | Job weighted average qtime, TotalQTS / JC |
ATAPH | average time queued PH ? |
ATQPH | average time queued PH ? |
AXF | PS weighted average xfactor, PSXFactor / PSRun |
GCAJobs | current active jobs |
GPHAvl | total processor hours- @see mstat_t.TotalPHAvailable -- processor hours available |
GPHDed | total processor hours dedicated - @see mstat_t.TotalPHDed |
GPHSuc | proc hours dedicated to jobs which have successfully completed |
GPHUtl | total processor hours used - @see mstat_t.TotalPHBusy |
IC | number of iterations contributing to record |
IMEM | idle mem |
INC | idle node count |
IPC | idle proc count |
MBP | maximum bypass count |
MinEffIteration | mininum efficiency iteration |
MQT | max queuetime of finished jobs, MAX(QT[J]) -- max queue wait time |
MXF | max xfactor |
RMPI | RM poll interval |
SCJC | successful completed job count |
StartTime | most recent time job started execution |
statInitTime | date stats were initialized |
TJA | total job accuracy |
TJC | total jobs finished during period, |J| -> JC |
TMSA | total memory-seconds (MB * sec) available |
TMSD | total memory-seconds (MB * sec) dedicated |
TMSU | total memory-seconds (MB * sec) utilized |
TNJA | total node weighted job accuracy |
TNJC | node weighted finished job count, SUM(NC[J]) -- total allocated node count (only updated on job completion) |
TPSD | per iteration -- total proc-seconds (proc * sec) dedicated |
TPSE | proc-seconds executed associated w/completed jobs (WallRunTime*PC) -- total proc-seconds (proc * sec) executed |
TPSR | proc-seconds requested associated w/completed jobs (WallReqTime*PC) -- total proc-seconds (proc * sec) requested |
TPSU | per iteration -- total proc-seconds (proc * sec) utilized |
TStartJC | S->StartJC (field 50) -- jobs started this interval |
TStartPC | S->StartPC -- proc-weighted jobs started this interval (avg job pc = TStartPC / TStartJC) |
TStartQT | S->StartQT -- total queuetime of jobs started this interval (avgqt = TStartQT / TStartJC) |
TStartXF | S->StartXF -- total xfactor of jobs started this interval (avgxf = TStartXF / TStartJC) |
TSubmitJC | S->SubmitJC -- jobs submitted this interval |
TSubmitPH | S->SubmitPH -- proc-hours submitted this interval (avg job ph = TSubmitPH / TSubmitJC) |
UPN | up node count |
Tags: mstats, statistics