Add first version of Kafka operator #3
Conversation
|
I'd be happy with just your review @maltesander but I've added @soenkeliebau as you requested. |
There was a problem hiding this comment.
I've looked through the code and overall am happy for it to go in. Left just a few minor comments.
One thing though, I get
" Reconciliation finished with an error, this should not happen, please file an issue err=ObjectNotFound { obj_ref: ObjectRef { kind: ErasedResourceState..." when running this. I think I remember you having that with the ZK operator as well and mentioning that it is not critical - but can't remember what it was.
| target/ | ||
| **/*.rs.bk | ||
|
|
||
| .idea/ |
There was a problem hiding this comment.
Super-Nit: this shouldn't be in the .gitignore that is checked in
There was a problem hiding this comment.
Because that is specific to your development environment and should be contained either in your global .gitignore file or in your local repositories .git/info/exclude file.
At least that's the way I was taught :)
But as I said, super-nit, feel free to leave as is.
There was a problem hiding this comment.
The reason I always include this is because I don't want anyone checking in their .idea even if they don't have it in their global .gitignore.
There was a problem hiding this comment.
Following that argument we should also include .eclipse, ~ files for vim, whatever vscode does ..
I totally understand where you are coming from and please, absolutely do merge like this!! This is really more of a philosophical discussion :)
| self.context.client.delete(pod).await?; | ||
| } | ||
| } | ||
| Ok(ReconcileFunctionAction::Continue) |
There was a problem hiding this comment.
If there were excess pods which we deleted, this was probably a cluster shrink. General question, should we requeue to enable performing checks whether the cluster has rebalanced after the shrink before taking any more actions?
Irrelevant in this case, as it is the last action anyway and there are no checks to perform..
There was a problem hiding this comment.
Not irrelevant. It's a good question.
Should we delete a single pod and then requeue or should we delete all of them?
In Kafka we probably want to take steps to rebalance before removing too many?
This is where you are the expert.
But before we have a full solution you're right and it might make sense to requeue after every deletion.
|
The reconciliation error you mentioned has been fixed in operator-rs today so that should hopefully not happen anymore |
No description provided.