[Linux-cluster] DLM Problem

[Patrick Caulfeld]

isplist at logicore.net wrote:
>> That means that something else is using port 21064 - the TCP port that
>> the DLM uses. If the DLM can't bind to its port then it cannot start.
> 
> What's weird is that I didn't change anything, just the storage, then needing 
> to get the cluster quorum so that I could set up gfs. 
> 
>> Use netstat -tap or lsof to find out what is using that port. If you
>> can't stop that particular application that is using it, then you'll
> 
> #lsof -i | grep 21064
> Reveals nothing.
> 
> # netstat -anp | grep 21064
> Seems to show port 21064 being available.
>  
> tcp	0	0 192.168.1.58:21064      0.0.0.0:*                   LISTEN      -
> tcp	0	0 192.168.1.58:6809        192.168.1.62:21064  ESTABLISHED -
> tcp	0	0 192.168.1.58:21064      192.168.1.63:32780  ESTABLISHED -
> tcp    0   0 192.168.1.58:21064	192.168.1.92:6809     ESTABLISHED -
> tcp	0	0 192.168.1.58:21064		192.168.1.62:6809     ESTABLISHED -
> tcp	0	0 192.168.1.58:21064      192.168.1.40:6809     ESTABLISHED -
> 
>

That's not showing available, that's showing it already in use. The port
6809 on the other end is very suspicious, like there is maybe some
confusion about cman & dlm ports going on. Check cluster.conf and your
startup scripts for port changing things. It's very unusual. Also check
that all the nodes are using the same configuration.

There certainly is some port conflict going on with that system. it
might be a bizarre cluster suite misconfiguration or it might be some
other application - it's hard to be sure.

Patrick