General question

Hello Peter,

after playing with enbd for a while now, I must admit I seem to have failed
understanding some major points, maybe someone can explain.

I built a raid1 config with two servers, one also acting as enbd-client for the
raid1-fs.
Now, I go and mount the raid1 and start bonnie on it, looping forever (so we
have some fs-action).
Everything is fine, the fs works.
Now I manually degrade the raid1 with setraidfaulty.
Everything is fine, the fs works flawlessly, bonnie just runs.
I "rebuild" with hotadd, which also works fine.

Since I'd like to use raid1 for being robust against network troubles, I try
unplugging the network between the two enbd-servers.
Now something interesting happens:
Bonnie stops right away (freezes), the raid1 in fact freezes. You cannot do
anything with it. You cannot even setraidfaulty the now-gone enbd-server
partition.
The situation is frozen until I re-enable network again. Then, bonnie just
continues, my also hanging "setraidfaulty" comes back and I can hotadd the
faulty enbd-partition again.

In fact, that is not what I had expected. My idea was to use a raid1 fs to be
able to _continue_ working on the fs even if one of the enbd-servers fails.
How can I achieve that?
-- 
Regards,
Stephan