FIO测试

  • 2014年4月4日
  • 5个答复
  • 1648年的观点
C
徽章 +5




是否有人使用FIO测试Nutanix，但没有诊断VM？
当我们测试Nutanix时，在主机上的处理器几乎已经装载了所有核心。
我们开始测试3.5.2.1，并使一些磁盘离线，我们将其返回并升级到3.5.3.1。
安装结果后的第一次是可以的（3350上的400 GB磁盘13K IOPS），但是第二天（所有数据都是冷的）有时我们有零计数器。例如0读/0写。Nutanix是幻灯片上的上部控制台。
它是正常的吗？
NCC检查告诉一切还可以。
为了在幻灯片上进行测试，我们使用100GB磁盘，因此我们应该在一个节点中。
FIO配置
> sudo fio read_config.ini
root@debian：〜＃cat test_vmtools.ini
[readtest]
blocksize = 4k
文件名=/dev/sdb
rw = randread
直接= 1
缓冲= 0
ioEngine = libaio
iodepth = 32

> sudo fio write_config.ini
root@debian：〜＃cat test_vmtools.ini
[writeTest]
blocksize = 4k
文件名=/dev/sdb
rw = randwrite
直接= 1
缓冲= 0
ioEngine = libaio
iodepth = 32

> sudo fio rw_config.ini
root@debian：〜＃cat test_vmtools.ini
[readtest]
blocksize = 4k
文件名=/dev/sdb
rw = randread
直接= 1
缓冲= 0
ioEngine = libaio
iodepth = 32
[writeTest]
blocksize = 4k
文件名=/dev/sdb
rw = randwrite
直接= 1
缓冲= 0
ioEngine = libaio
iodepth = 32

谁能帮我测试吗？
<\/a><\/p>
\n
\n
\nDoes anyone test Nutanix with FIO but not with diagnostics VM?
\nWhen we testing Nutanix our processors on host all cores are almost 100 % loaded.
\nWe started test 3.5.2.1 and get some disks maked offline, we return them and upgrade to 3.5.3.1.
\nFor the first time after install results is ok (13k IOPS for 400 Gb disk on 3350), but next day (all data is cold) we have afwul resuls with zero counters sometimes.For example 0 read\/0 write. Nutanix is upper consoles on slide.
\nDoes it normal resuls?
\nncc check tell that everything is ok.
\nFor test on slide we use 100Gb disk so we should be in one node.
\nFio config
\n>sudo fio read_config.ini
\nroot@debian:~# cat test_vmtools.ini
\n[readtest]
\nblocksize=4k
\nfilename=\/dev\/sdb
\nrw=randread
\ndirect=1
\nbuffered=0
\nioengine=libaio
\niodepth=32
\n
\n>sudo fio write_config.ini
\nroot@debian:~# cat test_vmtools.ini
\n[writetest]
\nblocksize=4k
\nfilename=\/dev\/sdb
\nrw=randwrite
\ndirect=1
\nbuffered=0
\nioengine=libaio
\niodepth=32
\n
\n>sudo fio rw_config.ini
\nroot@debian:~# cat test_vmtools.ini
\n[readtest]
\nblocksize=4k
\nfilename=\/dev\/sdb
\nrw=randread
\ndirect=1
\nbuffered=0
\nioengine=libaio
\niodepth=32
\n[writetest]
\nblocksize=4k
\nfilename=\/dev\/sdb
\nrw=randwrite
\ndirect=1
\nbuffered=0
\nioengine=libaio
\niodepth=32
\n
\nCould anyone help me with test?","quoteUsername":"comotoza","translations":{"Common":{"like":"Like","unlike":"Unlike"},"Forum":{"Quote":"Quote","Share":"Share"}}}">
该主题已关闭以供评论

5个答复

加里
秩
Userlevel 1
徽章 +4
你好，

我们经常在Nutanix性能实验室中经常在诊断VM之外运行FIO。在测试期间，您的IO为零的事实表明，集群中有问题。请打开一个支持案例，以便我们弄清楚发生了什么。

- 加里。
\n
\nWe run fio outside of diagnostics VM quite frequently in the Nutanix performance labs. The fact that you're getting Zero IO's during the test indicates that something is wrong in the cluster. Pleas e open a support case so that we can figure out what's going on.
\n
\n-Gary.","quoteUsername":"gary","translations":{"Common":{"like":"Like","unlike":"Unlike"},"Forum":{"Quote":"Quote","Share":"Share"}}}">
C
徽章 +5
NCC检查说一切都可以。

高性能持续5-10分钟，然后降级性能。
Diagnostics VM获得高性能，但仅运行5分钟...



来自诊断VM：
等待热缓存冲洗....................................................
运行测试“随机阅读iops” ...
开始fio_rand_read：星期六3月29:02:25 2014

49495 IOPS
end fio_rand_read：星期六3月29日22:04:08 2014

持续时间fio_rand_read：103秒
*******************************************************************************************

等待热缓存齐平...完成。
运行测试“随机写IOPS” ...
开始fio_rand_write：星期六3月29日22:04:11 2014

26559 IOPS
end fio_rand_write：星期六3月29日22:05:54 2014

持续时间fio_rand_write：103秒
\n
\nHigh Performance for 5-10 min, then degraded performance.
\nDiagnostics VMs get high performance but they run only fo 5 min...
\n

<\/a><\/p>
\n
\nFrom diagnostics VMs:
\nWaiting for the hot cache to flush ........................ done.
\nRunning test 'Random read IOPS' ...
\nBegin fio_rand_read: Sat Mar 29 22:02:25 2014
\n
\n49495 IOPS<\/b>
\nEnd fio_rand_read: Sat Mar 29 22:04:08 2014
\n
\nDuration fio_rand_read : 103 secs
\n*******************************************************************************
\n
\nWaiting for the hot cache to flush ... done.
\nRunning test 'Random write IOPS' ...
\nBegin fio_rand_write: Sat Mar 29 22:04:11 2014
\n
\n26559 IOPS<\/b>
\nEnd fio_rand_write: Sat Mar 29 22:05:54 2014
\n
\nDuration fio_rand_write : 103 secs","quoteUsername":"comotoza","translations":{"Common":{"like":"Like","unlike":"Unlike"},"Forum":{"Quote":"Quote","Share":"Share"}}}">

t
秩
UserLevel 2
徽章 +13
每个VM有多少个VMDK磁盘？如果只有一个，则您可能每VDISK都达到我们的OPLOG限制。

尝试创建多个磁盘，看看是否会延长降级之前的时间。

谢谢，
\n
\nTry creating multiple disks and see if that extends the time before degredation.
\n
\nThanks,","quoteUsername":"tonyholland00","translations":{"Common":{"like":"Like","unlike":"Unlike"},"Forum":{"Quote":"Quote","Share":"Share"}}}">
C
徽章 +5
有两个主要问题：
1.我们的测试不正确 - 在Nutanix中，我们不能仅将所有IOPS提供给一个VM，我们应该使用具有自己的磁盘的多个VM。
一个VM：2000读/500在VM和主机上写
两个VM：2500 READ/750在主机上写入VM和5000/1500
因此，如果我们要测试Nutanix存储性能，则应使用许多VM。
2.在测试期间，由于出现不记忆的问题，星际之门过程崩溃，所有请求都重定向到另一个CVM。
Nutanix工程师帮助我们使用Gflags。
代码中还没有解决方案，并且计划为3.5.5和4.0进行修复。
现在我们看不到0读和写。
前：



后：


我应该如何测试以了解每个VM可以获得的性能，而不是整个集群？
如果我们与VCLOUD主管一起对服务提供商进行攻击，我们如何保证IOPS每个VM并将其写入SLA？Scioc和每VM IOPS手动手动？
\n1. our test was incorrect - in Nutanix we cann't give all IOPS only to one VM, we should use multiple VMs with own disk.
\nOne VM: 2000 read\/500 write in VM and on hosts
\nTwo VMs:2500read\/750 write in VMs and 5000\/1500 on host
\nSo if we want to test Nutanix storage performance we should use many VMs.
\n2. during tests stargate process crash due to an issue with out of memory and all requests were redirected to another CVM.
\nNutanix engineer help us with gflags.
\nThere is no solution in the code yet and the fix is planned for 3.5.5 and 4.0.
\nNow we don't see 0 read and write.
\nbefore:

<\/a><\/p>
\n
\nafter:
\n

<\/a><\/p>
\nHow should I test to understand what performance I can get per VM not whole cluster?
\nIf we are takling about service provider with vCloud Director how we can guarantee IOPS per VM and write it in SLA? SCIOC and manually per VM IOPS?","quoteUsername":"comotoza","translations":{"Common":{"like":"Like","unlike":"Unlike"},"Forum":{"Quote":"Quote","Share":"Share"}}}">

dlink7
Userlevel 4
徽章 +19
您可以使用FIO或Iometer测试，如果您的用例为1VM，并且1 VDisk确保撞击出色的IO。没有系统的杰出IO为1。

由于Nutanix在每个节点上都有本地存储控制器，因此您可以使用磁盘限制Shares来进行Guartnee资源。
http://itbloodpressure.com/2013/12/02/data-locality-sql-vdi-on-the-same-nutanix-cluster/

另外，您也可以尝试ioanalyzer，它们具有一些预先配置的工作负载，不确定它是否与您的工作相匹配。
\n
\nSince Nutanix has local storage controllers on each node you can use disk limitsshares to guartnee resources.
\nhttp:\/\/itbloodpressure.com\/2013\/12\/02\/data-locality-sql-vdi-on-the-same-nutanix-cluster\/
\n
\nAlso you could try IOAnalyzer too, they have some pre-configured workloads, not sure if it matches what your doing or not though.","quoteUsername":"dlink7","translations":{"Common":{"like":"Like","unlike":"Unlike"},"Forum":{"Quote":"Quote","Share":"Share"}}}">
条款和条件
Learn more about our cookies.<\/a>","cookiepolicy.button":"Accept cookies","cookiepolicy.button.deny":"Deny all","cookiepolicy.link":"Cookie settings","cookiepolicy.modal.title":"Cookie settings","cookiepolicy.modal.content":"We use 3 different kinds of cookies. You can choose which cookies you want to accept. We need basic cookies to make this site work, therefore these are the minimum you can select. Learn more about our cookies.<\/a>","cookiepolicy.modal.level1":"Basic
Functional","cookiepolicy.modal.level2":"Normal
Functional + analytics","cookiepolicy.modal.level3":"Complete
Functional + analytics + social media + embedded videos"}}}">
Baidu