[Gluster-devel] Problems while upgrading from 1.2.x to 1.3.x

[NovA]

Hi everybody!

I'm trying to switch my GlusterFS unify storage from 1.2.tla184 to the
new 1.3.tla524... First attempt I've described earlier (
http://lists.gnu.org/archive/html/gluster-devel/2007-09/msg00004.html
), but I couldn't experiment on production system till now.

So, I have working setup for 1.2.х GlusterFS unify with a lot of data
scattered on 10 servers. I want to replace GlusterFS to 1.3.x and keep
all data. Is this possible?

Now I've just updated FUSE to 2.7.0-glfs5, reinstalled GlusterFS,
changing spec files appropriately (see previous post) and got several
strange behaviors:
1) Just after mounting GlusterFS first attempt to touch the mount
point lead to lock for several minutes. In particular, I've mount to
/home and "ls /", "ls /home" locks with no possibility to break. No
logs for client/server, "top" shows no activity...
2) If the strange locking is gone, it's possible to list files in
GlasterFS volume, but not all of them. While navigating the volume,
sometimes it shows contents of a directory after a while (when
namespace fills with entries, I believe), sometimes it shows nothing.
3) Any attempt to create file on GlusterFS volume fails with the error
"No space left on device" (tla472 crashes on this)...

My system is OpenSUSE 10.2 х86_64, kernel 2.6.8-0.5 (with fuse 2.7.0
glfs patched module).

Any recommendations will be helpful. I will try myself to switch off
any performance translators.

With best regards,
  Andrey