| 修改日期 | 修改人 | 备注 |
| 2018-03-07 16:57:10[当前版本] | 程辉 | 格式调整 |
| 2018-03-07 16:56:52 | 程辉 | 格式调整 |
| 2018-03-07 15:56:43 | 程辉 | 修改标题 |
| 2018-03-07 15:44:23 | 程辉 | CREAT |
Vdbench对存储进行64K大块顺序写测试时报如下信息：
13:13:23.670 Starting slave: ssh 10.5.113.184 -l root /inst/hs/vdbench50402/vdbench SlaveJvm -m 10.5.113.183 -n 10.5.113.184-23-180222-13.13.21.886 -l hd2-5 -p 5570 13:13:23.706 Starting slave: ssh 10.5.113.184 -l root /inst/hs/vdbench50402/vdbench SlaveJvm -m 10.5.113.183 -n 10.5.113.184-24-180222-13.13.21.886 -l hd2-6 -p 5570 13:13:23.736 Starting slave: ssh 10.5.113.184 -l root /inst/hs/vdbench50402/vdbench SlaveJvm -m 10.5.113.183 -n 10.5.113.184-25-180222-13.13.21.886 -l hd2-7 -p 5570 13:13:25.603 All slaves are now connected 13:13:26.599 Not every host is reporting Cpu statistics. Cpu reporting disabled 13:13:30.016 Starting RD=run1; I/O rate: Uncontrolled MAX; elapsed=90 warmup=10; For loops: None Feb 22, 2018 interval i/o MB/sec bytes read resp read write resp resp queue rate 1024**2 i/o pct time resp resp max stddev depth 13:13:31.585 1 53709.00 3356.81 65536 0.00 25.130 0.000 25.130 52.973 11.295 1413.6 13:13:32.267 2 57772.00 3610.75 65536 0.00 29.674 0.000 29.674 63.300 9.185 1730.4 13:13:33.303 3 70711.00 4419.44 65536 0.00 29.432 0.000 29.432 51.858 8.970 2112.9 13:13:34.347 4 71752.00 4484.50 65536 0.00 29.549 0.000 29.549 55.655 9.045 2150.0 13:13:35.383 5 71023.00 4438.94 65536 0.00 29.561 0.000 29.561 48.015 9.029 2129.5 13:13:36.449 6 71564.00 4472.75 65536 0.00 29.601 0.000 29.601 49.532 9.067 2149.9 13:13:37.273 7 56588.00 3536.75 65536 0.00 29.978 0.000 29.978 54.593 9.412 1719.9 13:13:38.323 8 70890.00 4430.63 65536 0.00 29.893 0.000 29.893 57.553 9.420 2150.1 13:13:39.372 9 71596.00 4474.75 65536 0.00 29.608 0.000 29.608 54.905 9.071 2150.0 13:13:40.423 10 71685.00 4480.31 65536 0.00 29.566 0.000 29.566 54.954 9.034 2150.0 13:13:41.262 11 57331.00 3583.19 65536 0.00 29.574 0.000 29.574 51.639 9.019 1720.0 13:15:12.565 Waiting for the last interval's statistics to be reported 13:15:12.565 Waiting for slave: hd1-1 13:15:12.565 Waiting for slave: hd1-4 13:15:12.565 Waiting for slave: hd1-6 13:15:12.565 Waiting for slave: hd1-7 13:15:15.115 Waiting for the last interval's statistics to be reported 13:15:15.115 Waiting for slave: hd1-1 13:15:15.115 Waiting for slave: hd1-4 13:15:15.115 Waiting for slave: hd1-6 13:15:15.115 Waiting for slave: hd1-7 13:15:17.665 Waiting for the last interval's statistics to be reported 13:15:17.665 Waiting for slave: hd1-1 13:15:17.665 Waiting for slave: hd1-4 13:15:17.665 Waiting for slave: hd1-6 13:15:17.665 Waiting for slave: hd1-7 |
然后vdbench运行卡死无输出。
1、怀疑两台压力机时间不同步导致，查看两台压力机时间配置，发现已配置NTP时钟同步。
2、分析vdbench输出，发现延时特大，怀疑跟vdbench压力设置过大有关，将vdbench测试配置文件修改如下（红色部分）：
hd=default,vdbench=/inst/hs/vdbench50402,user=root,shell=ssh hd=hd1,system=10.5.113.183 hd=hd2,system=10.5.113.184 sd=default,threads=32 //将该参数值从128改成32后问题解决。 sd=sd1,hd=hd1,lun=/dev/rhdisk2,size=250g,openflags=o_direct sd=sd2,hd=hd1,lun=/dev/rhdisk3,size=250g,openflags=o_direct sd=sd3,hd=hd1,lun=/dev/rhdisk4,size=250g,openflags=o_direct sd=sd4,hd=hd1,lun=/dev/rhdisk5,size=250g,openflags=o_direct sd=sd5,hd=hd1,lun=/dev/rhdisk6,size=250g,openflags=o_direct sd=sd6,hd=hd1,lun=/dev/rhdisk7,size=250g,openflags=o_direct sd=sd7,hd=hd1,lun=/dev/rhdisk8,size=250g,openflags=o_direct sd=sd8,hd=hd1,lun=/dev/rhdisk9,size=250g,openflags=o_direct sd=sd9,hd=hd2,lun=/dev/rhdisk2,size=250g,openflags=o_direct sd=sd10,hd=hd2,lun=/dev/rhdisk3,size=250g,openflags=o_direct sd=sd11,hd=hd2,lun=/dev/rhdisk4,size=250g,openflags=o_direct sd=sd12,hd=hd2,lun=/dev/rhdisk5,size=250g,openflags=o_direct sd=sd13,hd=hd2,lun=/dev/rhdisk6,size=250g,openflags=o_direct sd=sd14,hd=hd2,lun=/dev/rhdisk7,size=250g,openflags=o_direct sd=sd15,hd=hd2,lun=/dev/rhdisk8,size=250g,openflags=o_direct sd=sd16,hd=hd2,lun=/dev/rhdisk9,size=250g,openflags=o_direct wd=wd1,sd=sd*,xfersize=64k,rdpct=0,seekpct=0 rd=run1,wd=wd*,iorate=max,elapsed=90,interval=1,warmup=10 |
将控制并发压力的参数threads改小。
无
Vdbench,threads,Waiting for the last interval’s statistics to be reported
