User equivalence problem

[Chucks_k]

Hi all,
O/S: RHEL 4 64 bit
db: 10.2.0.3 2 node RAC (64 bit) with ASM

We have a working RAC solution however I cannot get Database control to work for it, after much investigation i found that user equiv is not set up correctly (i think) even though RAC is working. Below is a test:

On node2:

[oracle@mgdb2 proc]$ ssh mgdb1 "date;hostname"
Wed May 21 14:31:21 BST 2008
mgdb1.mps
[oracle@mgdb2 proc]$ ssh mgdb2 "date;hostname"
oracle@mgdb2's password:

On node 1:
[oracle@mgdb1 ~]$ ssh mgdb2 "date;hostname"
Wed May 21 14:35:36 BST 2008
mgdb2.mps
[oracle@mgdb1 ~]$ ssh mgdb1 "date;hostname"
Wed May 21 14:35:40 BST 2008
mgdb1.mps
[oracle@mgdb1 ~]$

Hence, as can be seen node 2 cannot authenticate itself (as its asking for the password). To further prove my point:

[oracle@mgdb2 proc]$ cluvfy comp nodecon -n all -verbose

Verifying node connectivity

WARNING:
User equivalence is not set for nodes:
        mgdb2
Verification will proceed with nodes:
        mgdb1

Checking node connectivity...


Interface information for node "mgdb1"
  Interface Name                  IP Address                      Subnet
  ------------------------------  ------------------------------  ----------------
  bond0                           xxx.xxx.xxx.xxx                    xxx.xxx.xxx.xxx
  bond0                           xxx.xxx.xxx.xxx                    xxx.xxx.xxx.xxx
  bond1                           xxx.xxx.xxx.xxx                    xxx.xxx.xxx.xxx


Check: Node connectivity of subnet "xxx.xxx.xxx.xxx"
  Source                          Destination                     Connected?
  ------------------------------  ------------------------------  --------------
  mgdb1:bond0                     mgdb1:bond0                     yes
Result: Node connectivity check passed for subnet "xxx.xxx.xxx.xxx" with node(s) mgdb

Check: Node connectivity of subnet "xxx.xxx.xxx.xxx"
Result: Node connectivity check passed for subnet "xxx.xxx.xxx.xxx" with node(s) mgd

Suitable interfaces for the private interconnect on subnet "xxx.xxx.xxx.xxx":
mgdb1 bond0:xxx.xxx.xxx.xxxbond0:xxx.xxx.xxx.xxx

Suitable interfaces for the private interconnect on subnet "xxx.xxx.xxx.xxx":
mgdb1 bond1:xxx.xxx.xxx.xxx
ERROR:
Could not find a suitable set of interfaces for VIPs.

Result: Node connectivity check failed.


Verification of node connectivity was unsuccessful on all the nodes.

On thing i must mention here is that the application only uses node 2. When firing up Dbconsole I can see 'no targets found' hence we can set nothing useful. Does the above mean we have to re-config ssh? or is there any other suitable method? I would also like to mention user/group names are the same on both nodes.

Thanks in advance,
Chucks

[LKBrwn_DBA]

You do need to re-config SSH and make sure ALL file system directories are the same and have same permissions on all nodes.

[Chucks_k]

Hi,

As this is a live production db we will have to oraganise downtime to do this. Is this a big job and is there a chance the db wont come back up?

I have also noticed authorized_keys on node 1 file size is double that of the one on node 2. Would i be able to copy this (seeing that connections work from node1) onto node2 (replacing the one on node 2) and make it work that way?

[LKBrwn_DBA]

No need for down time, the problem may be only the authorized-keys/known-hosts files.

Yes, you may copy the working authorized-keys over, just make sure node 1 key is included in the file.

PS: Also make sure the permissions for that file are: -rw-------