- Hello, considering bumping/skipping patch relea...
# rke2
m
• Hello, considering bumping/skipping patch releases. looking over the release treadmill and comparing 1.30.8 with 1.30.14, in-place upgrades seem insane? A lot of version churn in all the components. I have bumped versions up and down in pre-production, but now this is production and it's been 6 months. Should my policy be, never upgrade unless forced, and prefer migration, or are patch releases generally fine to jump around in? (we suck at migration lol). Thanks
c
what?
Why wouldn’t you want to keep things up to date. Do you like bugs and vulnerabilities?
m
Can't say that I do. Alright I'll get on the treadmill, but every chance I get I'm gonna lobby for a PaaS or SaaS... (just kidding, you know very well all the PaaSes and SaaSes are super mickey mouse products). I just want to know, can I jump around patch releases without too much worry? Or is it like, every change can possibly crash production?
c
Why would it do that?
m
I mean 1.30.14+rke2r4 has no assets, r2 then...
c
Do you avoid updating os packages for the same reason?
1.30 is eol. The last upstream patch release was months ago. You should be moving to 1.31 or newer.
m
we do update packages... until we are unsupported and have to migrate, even then we can't often migrate apps, so we have to run these old platforms for the old apps that there is no plan to update, only migrate off of or more realistically, re-implement.
• Rancher prime 2.9.11 (or bump to 12) support only v1.30.14 (Default) • v1.29.15 • v1.28.15 • v1.27.16
I'm not gonna bump to 12, but the k8s versions are the same anyways.
we have old clusters on v1.27.16 so we are stuck there until we can drag those forward.
c
If you're using prime then you can find packages and images for 1.30.14+rke2r4 in the prime artifacts .
m
a ha! thx.
c
But Kubernetes in general requires keeping up. 1.27 is what, a year without updates?
m
over 1000 days, it's rke1
maybe we should just go back to static HTML, or some pdfs, done.
we don't have SRE or devops staff, or even sysadmins won't touch application platforms, I myself am a re-purposed developer with grey beard.
the only reason we could even think about running k8s is because Rancher containerized it and gave it a slick web UI for the devs to log into.
thanks for the clue BTW... have a social to attend.