Wow 1.8.2 is out.... when did that happen 🤣 we're on 1.7.1 what the general experience of the upgrade? Pretty smooth this time, I've just started planning - I've updated QLCNIC qlcnic-ko drivers for 1.8.2 if anyone needs them for these ancient SPF+ cards (to match SLES 6.12.0-160000.36). I'm panning to give this go weekend after next - once again because we use the stack via Rancher the upgrade button was hidden from view, when I visited harvester direct it appeared and is now showing - however I'm not sure if its that or a broken longhorn volume that was blocking it from display....

We did 7 clusters and 5/7 were super smooth no issues. 2 of them had issues. One got stuck in a weird race condition with upgrading the managed fleet/helm charts and the CRDs got updated but not the others, backing it out was weird. The other was somehow AMD binaries got mixed in with our ARM nodes preloaded in a bad state and even canceling/rebooting/restarting the update didn't fix it. We don't really know how that happened, nor could I reproduce it later. But other than that it was super smooth sailing.

In our situation all the machines are x86, however we want to move/add arm later 😉 thanks 🙂

Yeah I have no idea what happened there. That cluster has gone through upgrades over the last year or so from like 1.5.1 wiht no issues.

@worried-state-78253 starting with Harvester v1.9.0 we've got a new

rancher/harvester-kernel-module-devel

image which is essentially the same thing as the nvidia driver toolkit, but without the nvidia entrypoint, which makes it bit nicer to use when building kernel modules for specific harvester releases going forwards. Hopefully that's useful for you with the qlcnic drivers. The docs just went up at docs.harvesterhci.io/v1.9/advanced/installing-additional-software#…

Thats cool, I'll get my build updated soon and give this a go.

sweet, please LMK if you run into any trouble