Cluster resource goes comatose error 20892

Question

Hi,
We've got a problem with 2 new cluster nodes, when trying to migrate a pol on to them the cluster resource goes comatose.
In the /var/opt/novell/log/ncs/ resource.load.out
we see an error 20892.
That is nss /poolact=res fails because the server thinks the resource is used by another cluster node. That is not the case.
I've read the TID about that error but there it was said that this issue is fixed with ncpserv later than Jan 10.
The node in question is OES2SP3 and ncpserv is 2.0.3-0.21 (May 2011).
If i do a "cluster status" it shows the resource running on another node, but wwhen I try and migrate the resource it goes comatose.
We are running a mixed cluster (4* NW6.5SP8, 1 OES2SP2 (which works), 2 OES2SP3 (don't work)).
Where to look next?
Regards,
Bettina

Answer 1

bkuhlmann wrote:
>
> Thank you for your responses. I solved the problem myself yesterday. The
> problem was that nss and afptcpd did not start correctly (skipped
> services on boot) and were restarted manually afterwards an that was the
> cause for the failing of the migration.
> I fixed the nss,afp problem by adjusting the settings in
> /etc/sysconfig/udev as mentioned in the docs.
> And after that the pool could be properly activated.
> Now the AFP binding leads to comatose resource but that is another
> problem (at the moment I think this binding is obsolete because in our
> testenvironment we dod not bind afp to the short resource name (AFPBIND
> ADD poolservername <ipaddress>) and our Macs use the DNS names of the
> servers anyway.)
> And, yes, we added LUNS and I fear we only "cluster leaved" the linux
> nodes and did not take them down...
>
> Regards,
> Bettina
> Bettina
>
>
I am trying to make sure I understand this. So you do not use the afpbind in
you cluster scripts and afp continues to work without issue? That just seems
strange that it would work. I will have to test this myself if I am
understanding you correctly.