Sysbench weirdness on SLES 10
2008-06-20 at 08:45 am Matthew YonkovitWanted to note some weirdness in using sysbench’s rnrd tests on SLES 10. Running some bencharks for a client on their new SAN this week and ran into an interesting issue. Whenever I run sysbench with the following:
time sysbench –num-threads=16 –test=fileio –file-total-size=20G –file-test-mode=rndrd run
The system CPU spikes to 90-100% of the machine and almost no io ends up going to the disk. Seems like I am missing something, used the same version on ubuntu & centos without the same issues. Tried local & san disk, the same pattern holds true.
avg-cpu: %user %nice %system %iowait %steal %idle 5.40 0.00 86.66 0.00 0.00 7.95 Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s rkB/s wkB/s avgrq-sz avgqu-sz await svctm %util sda 0.00 1.80 0.00 1.90 0.00 31.20 0.00 15.60 16.42 0.00 0.00 0.00 0.00 sdb 0.00 0.30 0.00 1.00 0.00 11.20 0.00 5.60 11.20 0.00 0.00 0.00 0.00 sdc 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
The following disk tests are fine however:
time sysbench –num-threads=16 –test=fileio –file-total-size=20G –file-test-mode=rndwr run
time sysbench –num-threads=16 –test=fileio –file-total-size=100G –file-test-mode=rndrw –file-rw-ratio=1 run
avg-cpu: %user %nice %system %iowait %steal %idle
0.01 0.00 6.94 5.78 0.00 87.25
Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s rkB/s wkB/s avgrq-sz avgqu-sz await svctm %util
sda 0.00 0.43 0.03 0.38 0.27 7.20 0.13 3.60 17.92 0.00 0.48 0.48 0.02
sdb 0.00 0.23 0.00 0.43 0.00 6.13 0.00 3.07 14.15 0.00 1.85 1.23 0.05
sdc 0.00 58386.72 0.08 556.98 0.67 470907.33 0.33 235453.67 845.34 67.17 120.58 1.15 64.19
So its only the read random test. Not sure what is going on yet, but I thought I would through this out here to remind me to test & figure this out later.