How do you serialize the resync? I don't see the "group" option any more, which is what I used to control priority and speed of the resync.<div><br></div><div>I'm on 8.3.0 now, and when I fire up a node that was down, _all_ devices start syncing, which only makes it take longer due to the underlying disk thrashing (is my guess).</div>
<div><br></div><div>--Sam<br><br><div class="gmail_quote">On Fri, Jan 23, 2009 at 10:44 AM, Lars Ellenberg <span dir="ltr"><<a href="mailto:lars.ellenberg@linbit.com">lars.ellenberg@linbit.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;"><div class="Ih2E3d">On Fri, Jan 23, 2009 at 05:59:47PM +0100, Bruno Voigt wrote:<br>
> Hi,<br>
><br>
> I'm looking for some figures:<br>
><br>
> What are the memory requirements/guidelines for DRBD resources?<br>
> What are the maxima for normal operation, sync, verify ?<br>
> Is it dependent on the size of the replicated device?<br>
> Does the requirements increase lineary with each resource ?<br>
><br>
> I'm playing with a Debian lenny/unstable system pair (HP DL360 G5, 8G RAM)<br>
> with linux 2.6.18.6 (or 2.6.26.1), xen 3.2.1, DRBD 8.3.0,<br>
> hosting 13 DomU - each with 2 DRBD-Devices (4G disk+512M swap)<br>
> allocated from LVM LVs each separately replicated with DRBD.<br>
><br>
> The Dom0 got a fixed amount of memory (1 G via grub menu.list<br>
> xenhopt=dom0_mem=1024M)<br>
> and I've seen the situation that the Out-of-memory killer strikes on the<br>
> node A<br>
> if all DRBD resources are synced (all at once) from node A to initially<br>
> connected node B.<br>
><br>
> Nothing else besides an small apache is running on the Dom0,<br>
> in disconnected mode there is over 600M available on node a.<br>
><br>
> If I sync the resources one by one the OOM can be prevented.<br>
><br>
> So I'm looking for a solid based way to size the RAM of the Dom0 ..<br>
<br>
</div>From: Lars Ellenberg<br>
Subject: Re: Summarizing DRBD memory usage<br>
Date: 2008-07-24 10:43:52 GMT<br>
<a href="http://thread.gmane.org/gmane.linux.network.drbd/15482/focus=15485" target="_blank">http://thread.gmane.org/gmane.linux.network.drbd/15482/focus=15485</a><br>
<br>
so in your situation,<br>
right, serialize the resync (probably a good idea anyways).<br>
and/or maybe decrease "max-buffers", which limits the number of<br>
in-flight pages.<br>
<br>
or give it plenty of ram.<br>
<font color="#888888"><br>
--<br>
: Lars Ellenberg<br>
: LINBIT | Your Way to High Availability<br>
: DRBD/HA support and consulting <a href="http://www.linbit.com" target="_blank">http://www.linbit.com</a><br>
<br>
DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.<br>
__<br>
please don't Cc me, but send to list -- I'm subscribed<br>
</font><div><div></div><div class="Wj3C7c">_______________________________________________<br>
drbd-user mailing list<br>
<a href="mailto:drbd-user@lists.linbit.com">drbd-user@lists.linbit.com</a><br>
<a href="http://lists.linbit.com/mailman/listinfo/drbd-user" target="_blank">http://lists.linbit.com/mailman/listinfo/drbd-user</a><br>
</div></div></blockquote></div><br></div>