I am running Vdbench502 in multi-host mode (4x AIX partitions) with a “random 8 KB 100% cache hits” workload. It happens that Vdbench runs OK until it reaches 216000 IOPS: at this point one of the slaves starts giving IO error and then Vdbench aborts. This is what I see during the run (at 208000 IOPS is the last good observation point, then when going up to 216000 IOPS it fails) . . .
00:34:15.006 Starting RD=Rwh100; I/O rate: 208000; elapsed=180; For loops: iorate=208000 rhpct=100.0 whpct=100.0 threads=16.0
Sep 29, 2015 interval i/o MB/sec bytes read resp resp resp
rate 1024**2 i/o pct time max stddev
00:35:15.410 1 208068.00 1625.53 8192 0.00 2.647 189.423 2.952
00:36:15.463 2 208029.60 1625.23 8192 0.00 2.209 114.004 1.401
00:37:15.324 3 208066.20 1625.52 8192 0.00 2.165 99.962 1.114
00:37:15.409 avg_2-3 208047.90 1625.37 8192 0.00 2.187 114.004 1.266
00:37:16.503 Waiting 180 seconds; requested by 'pause' parameter
00:40:21.019 Starting RD=Rwh100; I/O rate: 216000; elapsed=180; For loops: iorate=216000 rhpct=100.0 whpct=100.0 threads=16.0
Sep 29, 2015 interval i/o MB/sec bytes read resp resp resp
rate 1024**2 i/o pct time max stddev
00:41:21.326 1 212521.62 1660.33 8192 0.00 9.199 174.009 11.696
00:42:21.296 2 124826.83 975.21 8192 0.00 19.876 49090.433 433.864
00:42:22.901 P3-6: 00:42:22.743 op: write lun: /dev/rhdisk264 lba: 319488 0x0004E000 xfer: 8192 errno: 2 ENOENT No such file or directory. Seek error? File not open for output?
00:42:22.901 P3-2: 00:42:22.697 op: write lun: /dev/rhdisk264 lba: 1040384 0x000FE000 xfer: 8192 errno: 2 ENOENT No such file or directory. Seek error? File not open for output?
00:42:22.901 P3-7: 00:42:22.732 op: write lun: /dev/rhdisk264 lba: 1015808 0x000F8000 xfer: 8192 errno: 2 ENOENT No such file or directory. Seek error? File not open for output?
. . . the above error messages are just first three, but I do get a longer stream of messages like those ones and then Vdbench aborts.
Thank you
Gustavo