Call rebalance API once all pods are updated

radu-gheorghe commented 9 months ago

Timid attempt to fix #615

It works on my machine, it seems to fix the problem, but:

I have no tests. Yet. I have a feeling this is the hard part 🙂
I guess I'll need to update managed-updates.md a bit as well.
There are a couple of TODOes in my patch where I'm not sure if things are right - any feedback there would be great!
Last but not least, with or without my patch, I always got a DOWN shard at the end of my rolling update. That's a separate issue, right? Still, it's always reproducible in my environment, maybe due to the fact that I'm testing with some docs in my collection and I'm having my IDE (with Solr Operator) talk to Kubernetes via a port forward to the "common" service, which sometimes fails because the underlying pod gets restarted, so I get a few "connection refused" errors until I put it back. Still, it should be something that the Operator handles gracefully, IMO.

Any feedback is welcome, I'd be glad to push this forward!

radu-gheorghe commented 8 months ago

Looks good to me, too, thanks a lot @HoustonPutman!

I already had a look yesterday, too, and noticed you couldn't help yourself and fixed your 2) above 🙂

HoustonPutman commented 8 months ago

I already had a look yesterday, too, and noticed you couldn't help yourself and fixed your 2) above 🙂

Haha I remembered that we had the lovely readinessConditions that told us if the cloud was ephemeral, so it was very easy to do it!

apache / solr-operator