I've setup a few netapp's before but the clustering usually is straight forward....however I now have an issue with one of my FAS2020 at a customer site that won't failover.
They have one FAS2020, the controllers are named netapp1 and netapp2, these are configured with vif0 and vif1 respectively. Clustering is licensed and enabled, if I manually initiate a takeover it works, however if I disconnect the LAN cables on either netapp the failover doesn't take place...the still connected controller status show's the cluster partner as "UP". If I ping the partner hostname before I unplug the nics I get an "alive" message, after the disconnect of the nics the ping show's that it cannot communicate with the partner...so I can't see it being a dns issue.....
Has anyone come across this before or can point me in the right direction?
Solved! See The Solution
10 REPLIES 10
Re: Cluster failover issue
2011-09-27 01:44 AM
I think what you may be finding here is that disconnecting the LAN will not initiate a failover…
Doing a NIC fail does not constitute a controller failover, which is why the controllers don’t failover and show as up, the controllers don’t talk to each other over Ethernet, they talk to each other internally.
It is a bit of a pain I know…but as far as I am aware NIC down is not a fail.
What is it you are trying to do, maybe can suggest some alternative methods…
2011-09-27 01:49 AM
have a look at the options settings
and you need to include and entry into the /etc/rc file and the nd of the vif line nfo
thanks, I thought that may have been the case...so is there no way to initiate the failover if the nic goes down?
I have VM's on both controllers and want the failover to work if one switch goes down so that my vm's will continue to run.
So my setup is one FAS2020 connected to two stacked switches, one controller to one switch and the other to the other, the same goes for my Virtual hosts, one host to one switch and the other to the other, so effectively two hosts to two switches...to two controllers...I want to be able to failover should half the equipment fail.
I've seen this command on another thread here...."cf.takeover.on_network_interface_failure.policy" to "any_nic" would this do it?
at the end of the current line in the /etc/rc where it creates the vif just add the nfo, don't create a new line.
the failover happens also immediately, quick enough for nfs, iscsi applications to not be disrrupted.
if you want to discuss this skype me: michailm33
Re: Cluster failover issue
2011-09-27 02:03 AM
I think my company block Skype..
I see you have CYP at the end of your screen name...stand for Cyprus by any chance?
OK, just seen your profile!
I'm in Limassol, afgaristo bolli yia din voithesma