Add horizontal pod autoscaler

[Epicsteve2]

Adds horizontal pod autoscaler to the helm chart.
A horizontal pod autoscaler simple will simply scale the number of Trino pods based off of CPU usage.

[cla-bot]

Thank you for your pull request and welcome to our community. We require contributors to sign our Contributor License Agreement, and we don't seem to have you on file. In order for us to review and merge your code, please submit the signed CLA to cla@trino.io. For more information, see https://github.com/trinodb/cla.

[cla-bot]

Thank you for your pull request and welcome to our community. We require contributors to sign our Contributor License Agreement, and we don't seem to have you on file. In order for us to review and merge your code, please submit the signed CLA to cla@trino.io. For more information, see https://github.com/trinodb/cla.

[Epicsteve2]

Note: autoscaling/v2 is out, but it went General Availability for Kubernetes 1.23. I feel like that version is too new to include in this helm chart.

For reference, AWS doesn't even support Kubernetes 1.22 yet

[Epicsteve2]

CI failed because I forgot to commit my changes for removing query.max-total-memory-per-node. 😅

I squashed it to 1 commit, but let me know if it should be a separate commit.
I do notice that #18 is almost ready to merge now. Should I just wait for that?

[hashhar]

@Epicsteve2 The linked PR has been merged, can you rebase accordingly?

[hashhar]

LGTM % comment

[Epicsteve2]

Looks like I forgot to bump the chart minor version (0.4.0 -> 0.5.0), so I updated that as well.

[hashhar]

@Epicsteve2 Do most projects bump chart version for each change? I'd prefer to change them in batches when we explicitly decide to release some of those changes together.

[Epicsteve2]

@Epicsteve2 Do most projects bump chart version for each change? I'd prefer to change them in batches when we explicitly decide to release some of those changes together.

Oh, yeah good point. I'll revert the change again. My apologies 😅

[hashhar]

Thanks for the contribution @Epicsteve2. I plan to cut a new release once #19, #12 and #17 make in.

[laserninja]

@Epicsteve2 What happens if the CPU usage for a node dips below the threshold and the node is still serving a query? Will the query fail?

[Epicsteve2]

What happens if the CPU usage for a node dips below the threshold and the node is still serving a query? Will the query fail?

@laserninja
Yeah it kills the query when that happens 😭

Ideally, Trino wouldn’t kill queried pods when scaling down. I'm not sure if that's an issue being worked on for Trino, but it could be fixed in a future version of Trino

One way that could possibly be dealt with is to setup query retries for now. (I haven't tested this though 😅)