Description of problem:
[root@link-02 ~]# clvmd
clvmd could not connect to cluster manager
Consult syslog for more information

dlm: Can't bind to port 21064
dlm: cannot start lowcomms -98
Oct 18 21:17:00 link-08 kernel: dlm: Can't bind to port 21064
Oct 18 21:17:00 link-08 kernel: dlm: cannot start lowcomms -98
Oct 18 21:17:00 link-08 clvmd: Unable to create lockspace for CLVM: Address
already in use


From Patrick's email:
Two interesting facts:

- netcat (nc) was quite capable of binding to port 21064 even when DLM couldn't.
- unloading the DLM module and reloading fixed the problem.

Which leads me to the conclusion that it's some saved state in the DLM that's
getting in the way, possibly a cached nodeid or IP address.

Had the IP address of the node changed between closing down the previous cluster
instance and restarting it? (looking at the code that would certainly cause an
error, though I'm not sure it would be that one!).

I think this problem should be bugged and maybe we try to figure out just what
caused it. Certainly there is some state in the DLM that should be flushed when
it gets deactivated that might be a contributory cause.


Version-Release number of selected component (if applicable):
[root@link-02 ~]# clvmd -V
Cluster LVM daemon version: 2.01.14 (2005-08-04)
Protocol version:           0.2.1