Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On 09/25/2012 04:37 AM, Lars Ellenberg wrote:
> This may be a known issue (a distributed deadlock) which can happen if
> your "max-buffers" is too small wrt. your resync rate.
Interesting. We have FusionIO drives on our DRBD setup, and its full
sync rate can get ridiculously high, but that's not what we've observed
with verify. With verify and the way it follows the map block by block,
the rate is much slower than a full sync, and I've never had a full sync
hang, while verify has had problems on all of our hosts. Even a couple
VM hosts I tested.
We have this, if it means anything:
net {
cram-hmac-alg "sha1";
sndbuf-size 0;
max-buffers 20480;
unplug-watermark 20480;
}
syncer {
rate 150M;
verify-alg md5;
al-extents 3389;
}
I'm not entirely familiar with why these settings were used, since our
admin set them way back when. I've been using the defaults with my 8.4
testing. That may mean something. :)
It should be noted that we're currently on an old CentOS setup, and it's
using DRBD 8.3.10. :(
--
Shaun Thomas
OptionsHouse | 141 W. Jackson Blvd. | Suite 500 | Chicago IL, 60604
312-444-8534
sthomas at optionshouse.com
______________________________________________
See http://www.peak6.com/email_disclaimer/ for terms and conditions related to this email