I had a short time window recently to do some lab work, and it was time to do some firmware upgrades using the Dell OpenManage Integration for vCenter tool (is it just me or does this tool need badly an acronym?). I have showed you how to install and do host level updates here
. However I had a bunch of outstanding updates so I thought I would do it at the cluster level. I am using 2.3 of Dell’s software.
I had mixed success. I monitored the Job Queue / Log on the OpenManager Monitor page. It let me know of the hosts that had an issue. I then restarted them, and tried again. Did that twice in fact. Both times some stuff was updated and some not. I really had a lot of updates pending! In the morning it looks like two hosts were not complete.
So I am doing the updates on them at the host level. That is not seen as a job in the Job Queue (until it is finished it seems), but you can watch it in the Log. I should mention I put the host in maintenance mode, and in the Firmware wizard I said it should not reboot but use the next reboot to apply.
You can confirm when to reboot by seeing when in the Job Queue the job shows up and has successfully staged for a status. Then it is safe to restart.
My believe is that I can do the two hosts that had issues and were not successfully updated at the host level and all will be good. We will see. So far it looks good.
On one of my hosts that showed successfully staged, and I restarted, still showed scheduled updates when I ran the Firmware Update on the host. I did that to make sure everything was done but it turns out everything wasn’t – the BIOS update for example was still not done. Since it showed as scheduled I restarted again. That made the difference. Now good.
So it looks like much is done by a cluster level firmware upgrade, so that does save time. But I had to do a host level firmware upgrade on two hosts before everything was done. Important to remember that no outage for any of my services occurred. But more time on my part. So that is no good but also it is good that no outages in my applications occurred.
For future reference, even though this was not perfect, I would still do a cluster level update, and then any that did not work do at the host level. You still get everything updated, and with no outages, and no datacenter visits so it is a still a net gain, but hopefully it will get better over the next version or two – it really needs too.
As always, questions or comments are most appreciated.
=== END ===