Dear DRBD community,
We anticipate that sometime we'll have big write activity spikes on our
DRBD cluster, like someone writing a 100GB video work.
The network between the two main nodes is slow, so we may want to let
the secondary node fall behind, and/or we may use drbd-proxy.
I gave the new congestion management features in v8.3.10 a try without
drbd-proxy, and I humbly admit I don't get the whole picture, so please
let me ask a few questions.
First, is drbd-proxy required if we want to use the new congestion
settings in v3.8.10?
If it's not required, then I wonder if it would be useful in our setup.
With video data, compression won't help, and when I push large files to
the drbd resource, "free" tells me the overall buffers get as large as
7GB. What more would drbd-proxy do?
Another point that I don't get is how drbd uses the "congestion-fill"
parameter. If I set it low (like 1K), the secondary node falls behind as
soon as there is write activity, as I expected. But if I set it to 1G
and I rsync a 10GB file full of zeros to the FS on the resource, the
secondary node never falls behind and rsync slows down to the network
speed. Am I doing it wrong?
Finally, could someone please shed some light on the
"congestion-extents" parameter too? All the docs say is it has the same
limit as al-extents, but what is it for? Is it mandatory?
Our setup is composed of 3 nodes with stacked resources.
The low-level resources are replicated between nodes A and B through a
WiMax link capable of 5MBps.
The stacked resources are replicated to node S. A has a GigE connection
to S, but B has to use the WiMax link to reach S.
The underlying RAID array can handle ~300MBps sustained writes.
The servers have 8GB RAM, 7GB of which is free.
I'll send the drbd config files if someone wants.
Thanks in advance to anyone who can help make this clearer.
Lionel Sausin - Numérigrahe SARL.
_______________________________________________
drbd-user mailing list
[email protected]
http://lists.linbit.com/mailman/listinfo/drbd-user
