[BUG] 2.11.1 Take snapshot UI repo list can be slow

[sandervandegeijn]

What is the bug?

Screen.Recording.2023-12-27.at.09.43.39.mov

The list is populated after a while but it's unclear that the user needs to wait because opensearch is working.

What is the expected behavior?
Implement waiting indicator

What is your host/environment?
Default 2.11.1 images.

Do you have any screenshots?
N/A

Do you have any additional context?
N/A

[kamingleung]

@sandervandegeijn Initial thoughts – I think adding a waiting indicator make sense if it's trying to probe a list of repositories. The Select component supports displaying a loading indicator with disabled state.

I am also curious what are the reasons it takes a significant time to load the list. What types of repositories have you set up? That may help us thinking about edge cases when improving this (e.g. Should the loading time out? What error messages can we communicate to users?)

@bowenlan-amzn Would like to hear your insights on this.

[bowenlan-amzn]

The get repository is called when mounting this component

index-management-dashboards-plugin/public/pages/Snapshots/components/CreateSnapshotFlyout/CreateSnapshotFlyout.tsx

Line 74 in 5da75d1

await this.getRepos();

This call itself is kind of slow.

A better way is to pass down the repositories from this flyout's parent component Snapshots table

[sandervandegeijn]

@sandervandegeijn Initial thoughts – I think adding a waiting indicator make sense if it's trying to probe a list of repositories. The Select component supports displaying a loading indicator with disabled state.

I am also curious what are the reasons it takes a significant time to load the list. What types of repositories have you set up? That may help us thinking about edge cases when improving this (e.g. Should the loading time out? What error messages can we communicate to users?)

@bowenlan-amzn Would like to hear your insights on this.

That's a valid question, we have two implementations in use: Azure and S3 on an on premise S3 solution. In both cases getting the list of snapshots is slow to dog slow. This also surprises me somewhat, especially since we're peaking at 25gbit/s from the S3 solution we have. That one isn't slow at all (but it's a ton of data).

But there is no overall state it seems where you can just get the current snapshots, it seems the whole repo is scanned each time. Having some kind of state would be beneficial (but introduces possible synchronisation problems). This would make the flow a lot faster I suspect.