Considerations on how to change IP address and hostname of an ESA node in cluster

[amelo@]

The cluster has four ESAs at the moment (three are M190, one is M170) running version 10.0.3-004.

I want to remove the M170 from the cluster and change the IP/hostname of one of the M190s to the IP/hostname of the M170 that was removed. The M170 is not connect to the SMA as it is part of the backup scenario (the M190 taking its place would not be connect to the SMA either.)

Can I do this change of IP/hostname while the node is in the cluster?

Are there any consideration I should have in mind?

Thanks,

[Ken Stieers]

To save your sanity I would eject both the m190 and the m170 from the cluster. turn off the m170, re ip the 190. Make sure to update your DNS records and then clear the DNS cache on all of the other esa's in the cluster. Then join the 190 back to the cluster.

[Mathew Huynh · Accepted solution]

To add to this.
You may see hostkey errors if you're swapping IPs to a new device when communicating in a cluster.

As Ken stated, by ejecting both the C170 and C190 in question, it should rid any of the hostkey exchanges - then you would be able to power off the C170, switch the IPs over on the C190 and commit changes.

Then re-join that C190 to your existing cluster.
If you choose -not- to remove the C190 off the cluster; you'll need to change the IPs on the C190 (after C170 is removed off the cluster, then switched off to avoid IP conflict).
Change the clusterconfig > communication to use the right IP for the right device
Do your hostkey scans again, so on the devices in the network - clear off any old hostkeys associated to that IP (CLI > logconfig > hostkeyconfig) and re-scan to store the newer keys to avoid a SSH key error.

(SSH key error will likely be more prominent in version 11.0+).

Regards,
Matthew

[amelo@]

Thanks for the suggestions. My fist thought was removing both from cluster, tuning off and switching IP/hostname. I'll try that.

Thanks for your help, 

[Mathew Huynh]

Hello Amelo@,

Let us know how you go :).

Cheers,

Matthew

[amelo@]

Ejecting both, making the changes and then joining cluster again worked without a hitch.
Thanks for the help.