<div class="gmail_quote">Am 13. April 2012 11:42 schrieb Brian Candler <span dir="ltr"><<a href="mailto:B.Candler@pobox.com">B.Candler@pobox.com</a>></span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im">On Fri, Apr 13, 2012 at 11:25:58AM +0200, Philip wrote:<br>
> Sometimes the io-latency skyrockets and there is no<br>
> access possible on the raid for >10 seconds.<br>
<br>
</div>Have you checked<br>
<a href="http://community.gluster.org/a/linux-kernel-tuning-for-glusterfs/" target="_blank">http://community.gluster.org/a/linux-kernel-tuning-for-glusterfs/</a><br>
?<br>
<br>
If you have a large amount of RAM and a lot of writes, maybe you're<br>
accumulating large amounts of dirty data which is then being synchronously<br>
flushed.<br></blockquote><div><br>I indeed have lots of ram but I've disabled all write access for a few hours to check if this issue is write related. These latency spikes did also happen when there were no writes at all.<br>
</div></div><br>I haven't changed these values. They were at<br>vm.dirty_background_ratio = 10<br>vm.dirty_ratio = 20<br><br>I've changed it to<br>vm.dirty_background_ratio = 2<br>vm.dirty_ratio = 10<br>