I’m getting into ScaleIO (Software Only) so am still quite new to the ecosystem and management commands, but having just gone through an upgrade cycle to the 2.5 release (from a later version than 126.96.36.199) I thought I’d note down my experience from doing a ScaleIO 2.5 upgrade and some things that might be useful to anyone else who does this.
Please note that this is just me experience and I am not to be considered a guru in all things ScaleIO. I’m happy to be corrected and educated 🙂
Ensure these are accepted before you begin otherwise you’ll have to accept them and restart. Not a big deal, but just one of those nice things for a smooth process 🙂
After extracting the software upgrade there are two folders – SLES11_3 and SLES11_3.SVM. Installation Manager will show you which one you need to upload:
Mine were identical on every node and I saw no problems but I would expect there to be errors if not.
Upgrade the SDCs automatically
I don’t know whether this box should be ticked. The PDF said to tick it, and a VxRack guide said not. I would guess that this should install the vibs ready for the ESX restart, but I don’t know for sure. As it would need ESX root passwords and SSH enabled I would hope that we would get prompted.
ScaleIO GUI Tools
The GUI Tools for 188.8.131.52 appeared to work with 2.5. The guide doesn’t say when to upgrade them – I did after the install completed but before upgrading the SDCs (Hosts).
Enabling SSH on ESXi hosts
This is so much easier via the HTML5 interface !
Instead of using the ScaleIO plugin and enabling SSH / root access to upload the VIBS – why not upload the zip file to Update Manager, configure a baseline and remediate each host ? Way easier !
The power down sequence for a host with an SDS is:
- Maintenance mode the SDS node
- Power down the SDS node
- Maintenance mode the ESXi host
Then we can remediate. Reverse the order to bring back online.
Switch the Primary/Master MDM
You will need to run the commands:
scli –login –username <username>
(you will then be prompted for a password)
scli –switch_mdm_ownership –new_mdm_target_id <ID>
I got the ID from the scli –query_cluster command, where it must be one of the slave MDMs.
I love how fast this is 🙂
Overall, this was relatively painless upgrade with so much potential to automate – especially the ESXi section. But that’s for another post !