* Jon LaBadie <[email protected]> [20100819 18:04]: > On Thu, Aug 19, 2010 at 03:27:07PM -0400, Jean-Francois Malouin wrote: > > Hi, > > > > I know that a little while ago I reported this problem but I still > > can't get my head around it: ever since I upgraded to amanda-3.1.x the > > avg tape stats in amreport are bogus. I know for a fact that the > > holddisk io read stats (from munin) --as this is when the taper starts > > pushing stuff out of the holddisk to the tape-- are consistently above > > 100MBs for the run yet amreport says ~14MBs. This lead me to suspect > > that the starting time for the taper to start to write the first DLE > > is wrongly set or parsed or something along this. > > > > Parts Taped 33 30 3 (1:3) > > Avg Tp Write Rate (k/s) 14644.4 14648.0 256.4 > > > > DUMP SUMMARY: > > DUMPER STATS > > TAPER STATS > > HOSTNAME DISK L ORIG-MB OUT-MB COMP% MMM:SS KB/s MMM:SS > > KB/s > > ------------------------- -------------------------------------- > > -------------- > > gaspar /opt/amanda 0 542 542 -- 0:34 16401.1 0:10 > > 55471.0 > > gertrude /opt/amanda 0 484 484 -- 0:16 31234.0 809:11 > > 10.2 > > tutor -aid/bicadmin1 0 86061 86061 -- 72:15 20329.5 12:27 > > 117973.9 > > tutor /raid/catlim1 0 715398 715398 -- 736:59 16566.8 113:14 > > 107825.6 > > tutor /raid/data1 1 1 1 -- 0:03 281.0 0:05 > > 190.0 > > tutor /raid/data2 1 2 2 -- 0:04 354.9 0:05 > > 308.0 > > tutor /raid/data3 1 1 1 -- 0:03 334.8 0:04 > > 275.0 > > tutor /raid/data4 0 0 0 -- 0:00 180.7 0:04 > > 2.5 > > tutor /raid/data5 0 0 0 -- 0:00 177.1 0:05 > > 2.0 > > tutor /raid/data6 0 0 0 -- 0:00 183.0 0:05 > > 2.0 > > tutor /raid/data7 0 0 0 -- 0:00 178.9 0:04 > > 2.5 > > tutor /raid/data8 0 0 0 -- 0:00 182.4 0:05 > > 2.0 > > tutor /raid/proj0 0 0 0 -- 0:00 180.3 0:05 > > 2.0 > > wart /opt/amanda 0 592 592 -- 0:18 34354.8 0:11 > > 55100.9 > > > > I think the calculation is correct (total column 4 / total column 9 is > 14.3MB/s). > > But you're observation of taping at >100MB/sec is also correct :) > > Note your client "gertrude". That one DLE is throwing the average to > hell<-<-ck. > Of your total data taped, 803,077MB, gertrude is only contributing 484MB. > However her time to tape is 809 minutes out of 936 minutes total. So her > 0.5% of the data taped takes 86.4% of the time to tape. > > Leaving gertrude out of the calculation I calculate the others tape at > 105.3MB/s. > > Why is gertrude so slow?
I don't think it's slow. The logfile for this run shows: START taper datestamp 20100818201101 label av48-2_down_Q00048 tape 1 SUCCESS dumper tutor /raid/data2 20100818201101 1 [sec 4.339 kb 1540 kps 354.9 orig-kb 1540] INFO dumper pid-done 1235 SUCCESS chunker tutor /raid/data2 20100818201101 1 [sec 4.438 kb 1540 kps 354.2] INFO chunker pid-done 1234 STATS driver estimate tutor /raid/data2 20100818201101 1 [sec 6 nkb 1572 ckb 1600 kps 235] PART taper av48-2_down_Q00048 1 gertrude /opt/amanda 20100818201101 1/-1 0 [sec 7.303252 kb 495980 kps 67912.213876 orig-kb 495980] DONE taper gertrude /opt/amanda 20100818201101 1 0 [sec 48551.000000 kb 495980 kps 10.215650 orig-kb 495980] So it took 7sec to write the part to tape and the DONE line shows 48551.000000 sec. Where is this number coming from? It took 13hours to write the header, the part and the filemark?? jf > > jl > -- > Jon H. LaBadie [email protected] > JG Computing > 12027 Creekbend Drive (703) 787-0884 > Reston, VA 20194 (703) 787-0922 (fax)
