<br><br><div class="gmail_quote">On Wed, Aug 25, 2010 at 8:54 AM, Dietmar Maurer <span dir="ltr"><<a href="mailto:dietmar@proxmox.com">dietmar@proxmox.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Hi Kir,<br>
<div class="im"><br>
> These numbers very much resemble fsync() rate with write cache enabled<br>
> (~1000/sec) and disabled (50-70/sec).<br>
> check write cache settings with hdparm + check whether you have barrier<br>
> mount option on ext3.<br>
<br>
</div>I use exactly the same system. The only thing I change is the kernel. kernel with<br>
OpenVZ patches is factor 20 slower!<br>
<br>
That is why I ask if you can reproduce the problem?<br></blockquote><div><br></div><div>poor performance here Dietmar, whit a SAS 10Krpm drive:</div><div><br></div><div>m2:~# sysbench --test=fileio --file-num=1 --file-total-size=20G --file-fsync-all=on --file-test-mode=seqrewr --max-time=100 --file-block-size=4096 --max-requests=0 run</div>
<div>sysbench 0.4.12: multi-threaded system evaluation benchmark</div><div><br></div><div>Running the test with following options:</div><div>Number of threads: 1</div><div><br></div><div>Extra file open flags: 0</div><div>
1 files, 20Gb each</div><div>20Gb total file size</div><div>Block size 4Kb</div><div>Calling fsync() after each write operation.</div><div>Using synchronous I/O mode</div><div>Doing sequential rewrite test</div><div>Threads started!</div>
<div>Time limit exceeded, exiting...</div><div>Done.</div><div><br></div><div>Operations performed: 0 Read, 3323 Write, 3323 Other = 6646 Total</div><div>Read 0b Written 12.98Mb Total transferred 12.98Mb (132.88Kb/sec)</div>
<div> 33.22 Requests/sec executed</div><div><br></div><div>Test execution summary:</div><div> total time: 100.0286s</div><div> total number of events: 3323</div><div> total time taken by event execution: 100.0146</div>
<div> per-request statistics:</div><div> min: 24.22ms</div><div> avg: 30.10ms</div><div> max: 89.84ms</div>
<div> approx. 95 percentile: 33.97ms</div><div><br></div><div>Threads fairness:</div><div> events (avg/stddev): 3323.0000/0.00</div><div> execution time (avg/stddev): 100.0146/0.00</div>
<div><br></div><div>m2:~# uname -a</div><div>Linux m2 2.6.32.15 #1 SMP Fri Jul 16 20:17:30 CEST 2010 x86_64 GNU/Linux</div><div><br></div><div>I think it's budarin.1 kernel.</div><div><br></div><div><br></div><div> </div>
</div>-- <br>Marc<br>