[Linux-cluster] problem with rejoining a node

[Javi Polo]

On Aug/09/2005, Lon Hohberger wrote:

> When using SAN fencing, the easiest set of steps to follow to ensure
> data integrity + correct operation:
> (a) power off machine
> (b) un-fence at the SAN level
> (c) power on machine

I've made a script that, prior to starting any of the cluster
infrastructure, enables his SAN port.

I can then join the cluster, but when I try to join the fence, it locks
up there ... :

gfstest1:~# cman_tool services
Service          Name                              GID LID State     Code
Fence Domain:    "default"                           0   2 join      S-1,80,3
[]

gfstest1:~# cman_tool nodes
Node  Votes Exp Sts  Name
   1    1    3   M   gfstest1
   2    1    3   M   gfstest2
   3    1    3   M   gfstest3
gfstest1:~# cman_tool status
Protocol version: 4.0.1
Config version: 9
Cluster name: test_cluster
Cluster ID: 61876
Membership state: Cluster-Member
Nodes: 3
Expected_votes: 3
Total_votes: 3
Quorum: 2   
Active subsystems: 1
Node addresses: 192.168.0.1

gfstest1:~# 

from other nodes, I see it as recovering:
gfstest2:/etc/init.d# cman_tool services
Service          Name                              GID LID State     Code
Fence Domain:    "default"                           1   2 recover 2 -
[2 3]

what happent?

-- 
Javier Polo @ Datagrama
902 136 126