takeover - not sure why?

[kongfranon]

Hi,

Hopefully posting this in right forum.

I have two netapp, and one of them tookover the other.  I think it happen a few months back, I looked at all the logs and do not see any messages about it taking over.

Is there anyway to find out why?  Should I just do a giveback?

Thanks

[watan]

I would open a support ticket with NGS to ensure the failed node is ok.  You don't want to try doing a giveback and cause other problems without diagnosing the failed node.  If it did a failover there is probably a good reason why it did so would strongly recommend having NGS take a look at the logs to be sure you don't have any failed loops, h/w, etc.

[kongfranon]

Thanks,

out of curisoity when a node is taken over you cannot telnet into it?  i wanted to take a look at the logs on that node and it would not let me connect at all, only the one that tookitover

[chriszurich]

Once the secondary controller has taken over the failed controller you can only telnet/ssh to the secondary.  You can however execute commands on the "taken over" controller by running partner from the console.

[aborzenkov]

Additionally if node exports root volume using CIFS/NFS, it is always possible to access /etc/messages and other files even if it is taken over.

[matej]

Have you done DataONTAP upgrade lattely? In latest version of DoT you should increase takeover detection time from 10 seconds to 15 seconds. Please take a look at opitons cf.takeover.detection.seconds.

And of course - pass the logs to NGS.

Matej

[watan]

Unfortunately you cannot telnet into the node that has been taken over.  Its essentially in either "waiting for giveback" mode or its completely down.  If you have access to MYautosupport you should be able to take a look at all the logs avaialble for both nodes.  Again , I would strongly recommend getting NGS involved if you haven't done so already.   Good luck!

[kongfranon]

Thanks everyone for the quick answers.  I have a call open and waiting to find out what support says.