Hi Admins,
We have a 4 Node Novell Cluster running on OES2 Linux. About 1 year ago we migrated the cluster nodes from NetWare to Linux, at this time to OES2SP1.
Now I patched one node after an other to the newest supportpack and I noticed, that the Master_IP_Address Resource often goes comatose, when I try to load the ressource on a newly patched server. Yesterday I got the last node with first step of patching process to OES2SP2 and now the Master_IP_Address keeps offline or comatose.
I found this in the log /var/opt/novell/log/ncs/Master_IP_Address_Resource.load.out:
+ exit_on_error ncpcon bind --ncpservername=ODNWCL --ipaddress=172.20.11.110
+ eval ncpcon bind --ncpservername=ODNWCL --ipaddress=172.20.11.110
++ ncpcon bind --ncpservername=ODNWCL --ipaddress=172.20.11.110
... Executing " bind"
... FAILED completion [elapsed time = 297 usecs]
+ rc=1
It seems to be an issue with ncpcon bind .. or slp or something protocol specific.
After some try and errors, some hours reading documentation and Forum Post I found the reason and I want to post it here, maybe someone out there has the same issue.
Novell changed the syntax of the load/unload scripts. Instead of spaces there are underscores:
Before: add secondary ipaddress 172.20.11.110
After SP2: add_secondary_ipaddress 172.20.11.110
This is an very evil detail change with no backwards compatibility ... I'm not amused, Novell ..
Have a look at the load script--> /var/opt/novell/ncs # cat Master_IP_Address_Resource.load
You may edit this file, but it wont have any effect. Use an LDAP Browser/Editor and edit the Master_IP Address_Ressource object directly. There you will find the attributes "Unload and Load Script" and replace the spaces with underscores. You may use iManager (Modify Object) too, but I haven't checked it out.
Then you can load your Master IP Ressource (it should be offline, take it offline in case it shows comatose) : cluster online Master_IP_Address_Resource
Additional Info:
---------------------
If you want to force access to the Cluster via iManager while the Master IP Address stays comatose, here are the commandlines, invoke them at the node, where the ressource is comatose:
ip addr add 172.20.11.110/16 broadcast 172.20.255.255 dev eth0
ncpcon bind --ncpservername=ODNWCL --ipaddress=172.20.11.110
[..Now you can view the cluster via iManager plugin .. I got only ReadOnly access .. ]
ncpcon unbind --ncpservername=ODNWCL --ipaddress=172.20.11.110
ip addr del 172.20.11.110/16 dev eth0
Good luck
Hardy, Germany
We have a 4 Node Novell Cluster running on OES2 Linux. About 1 year ago we migrated the cluster nodes from NetWare to Linux, at this time to OES2SP1.
Now I patched one node after an other to the newest supportpack and I noticed, that the Master_IP_Address Resource often goes comatose, when I try to load the ressource on a newly patched server. Yesterday I got the last node with first step of patching process to OES2SP2 and now the Master_IP_Address keeps offline or comatose.
I found this in the log /var/opt/novell/log/ncs/Master_IP_Address_Resource.load.out:
+ exit_on_error ncpcon bind --ncpservername=ODNWCL --ipaddress=172.20.11.110
+ eval ncpcon bind --ncpservername=ODNWCL --ipaddress=172.20.11.110
++ ncpcon bind --ncpservername=ODNWCL --ipaddress=172.20.11.110
... Executing " bind"
... FAILED completion [elapsed time = 297 usecs]
+ rc=1
It seems to be an issue with ncpcon bind .. or slp or something protocol specific.
After some try and errors, some hours reading documentation and Forum Post I found the reason and I want to post it here, maybe someone out there has the same issue.
Novell changed the syntax of the load/unload scripts. Instead of spaces there are underscores:
Before: add secondary ipaddress 172.20.11.110
After SP2: add_secondary_ipaddress 172.20.11.110
This is an very evil detail change with no backwards compatibility ... I'm not amused, Novell ..
Have a look at the load script--> /var/opt/novell/ncs # cat Master_IP_Address_Resource.load
You may edit this file, but it wont have any effect. Use an LDAP Browser/Editor and edit the Master_IP Address_Ressource object directly. There you will find the attributes "Unload and Load Script" and replace the spaces with underscores. You may use iManager (Modify Object) too, but I haven't checked it out.
Then you can load your Master IP Ressource (it should be offline, take it offline in case it shows comatose) : cluster online Master_IP_Address_Resource
Additional Info:
---------------------
If you want to force access to the Cluster via iManager while the Master IP Address stays comatose, here are the commandlines, invoke them at the node, where the ressource is comatose:
ip addr add 172.20.11.110/16 broadcast 172.20.255.255 dev eth0
ncpcon bind --ncpservername=ODNWCL --ipaddress=172.20.11.110
[..Now you can view the cluster via iManager plugin .. I got only ReadOnly access .. ]
ncpcon unbind --ncpservername=ODNWCL --ipaddress=172.20.11.110
ip addr del 172.20.11.110/16 dev eth0
Good luck
Hardy, Germany