scheduled job to upgrade all workers?

[nycalex]

good morning

i was thinking... what if i create a on-demand (manual execution) job to run as root a script "/opt/cronicle/bin/control.sh upgrade" on all worker nodes?

and run it detached, to be safe...

Would it work or a bad idea?

Thanks!!!

[jhuckaby]

LOL! That's funny.

I suppose if you ran it detached it might work, but I've never tried such a thing before. There may be unforeseen issues.

I definitely wouldn't attempt this on a production cluster.

[nycalex]

What are the possible reasons it's not a production-grade idea?

An ability to upgrade an entire cluster is a huge plus

[jhuckaby]

I dunno, I just haven't tried it before. I honestly don't know what will happen. Maybe nothing? Maybe it'll work great?

The risk is that it's updating itself from inside itself, while it is running a job, itself. It just seems like asking for trouble, is all.

Do you have some test servers you can try it on, that aren't running your live production jobs?

[nycalex]

I am just playing in a vm environment, i can try next time there's an update available.

[jhuckaby]

Cool!

An ability to upgrade an entire cluster is a huge plus

I totally agree this would be a great built-in feature. I'm working on something like this for v2 (due out later this year).