[Upgrade Assistant] Better handling of closed indices

[jloleysens]

Summary

Fix #41572

We use the _cluster/state endpoint to determine whether an index is open or closed. We inform the user in the UI but we still allow them to request a reindex with the understanding that the UA will open, reindex, alias and close the index again.

How to test

• Replace line 57 with the snippet below in x-pack/plugins/upgrade_assistant/server/lib/es_migration_apis.ts.
• Make sure that indices test1, test2 and test3 actually exist in the cluster. (e.g. in Console PUT test1 etc).
• Close test1. POST /test1/_close (more information on this API here)
• Navigate to the Upgrade Assistant "Indices" tab and expand the accordions until you find test1. There should be an alert icon next to the button with some more information displayed on hover.
• Follow the steps to start the reindex process. Once reindexing has completed, go back to Console and use GET _cluster/state/metadata/test1 and check that the state key is still close.
• Attempt reindexing test2, this should work as before and leave the index in open state (use same step as above but for index test2 to verify).

Release note

TODO

Additional information

Snippet:

      indices: indices.concat([
      {
        level: 'none',
        message: 'none level test',
        url: 'test/url',
        details: 'test details',
        index: 'test1',
        reindex: true,
        needsDefaultFields: false,
        blockerForReindexing: 'index-closed', // Faking that test1 is closed!
      },
      {
        level: 'info',
        message: 'info level test',
        url: 'test/url',
        details: 'test details',
        index: 'test2',
        reindex: true,
        needsDefaultFields: false,
      },
      {
        level: 'warning',
        message: 'warning level test2',
        url: 'test/url',
        details: 'test details',
        index: 'test3',
        reindex: true,
        needsDefaultFields: false,
      },
      {
        level: 'critical',
        message: 'metricbeat-1 successful fix',
        url: 'test/url',
        details: 'test details',
        index: 'metricbeat-1',
        reindex: false,
        needsDefaultFields: true,
      },
      {
        level: 'critical',
        message: 'metricbeat-2 failed fix',
        url: 'test/url',
        details: 'test details',
        index: 'metricbeat-2',
        reindex: false,
        needsDefaultFields: true,
      },
    ]),

Screenshots

Reindex Button state when index is closed

Flyout state (another callout, see emphasis)

Checklist

Delete any items that are not applicable to this PR.

• Any text added follows EUI's writing guidelines, uses sentence case text and includes i18n support
• Unit or functional tests were updated or added to match the most common scenarios

For maintainers

• This was checked for breaking API changes and was labeled appropriately

[elasticmachine]

Pinging @elastic/es-ui (Team:Elasticsearch UI)

[jloleysens]

@elasticmachine merge upstream

[jloleysens]

@elasticmachine merge upstream

[jloleysens]

@elasticmachine merge upstream

[alisonelizabeth]

@jloleysens great work! Tested locally with stubbed data.

I had a few UX suggestions/questions. Let me know what you think.

• The alert icon feels a little weird to me. At first, I was thinking the icon should be next to the index name, but I understand why you put it next to the button. Another thought I had was to change the reindex button color to warning in this scenario. Thoughts? You should still be able to keep the tooltip.

I also was not expecting the icon to be a link to documentation until I read through the code, but this is probably a user error on my part.

• Would it be beneficial to include the same warning on the reindex flyout?

• Do the "Reindexing process" steps on the reindex flyout need to be modified at all?

[alisonelizabeth]

Is it worth having a quick conversation with an ES engineer to confirm this is the correct approach?

[jloleysens]

I spoke with @DaveCTurner about this and we concluded that _cluster/state is probably the best place to get information about closed indices.

The addition of https://github.com/jloleysens/kibana/blob/d73316a2bb5e9d0436ba9d324a678e02b344cb6c/x-pack/test/upgrade_assistant_integration/upgrade_assistant/status.ts#L28

Should keep us on top of any changes that may happen to that response.

[alisonelizabeth]

LGTM. Thanks for makings those changes! I left one copy suggestion, but nothing blocking.

[jloleysens]

@elasticmachine merge upstream

[kibanamachine]

💚 Build Succeeded

• continuous-integration/kibana-ci/pull-request
• Commit: 1af9b34

History

• 💛 Build #31619 was flaky 42dbd5a
• 💛 Build #31471 was flaky 64b8e92
• 💔 Build #31419 failed 97d6d72
• 💔 Build #31404 failed a0e0ad4
• 💔 Build #31394 failed 15866c9

To update your PR or re-run it, just comment with:
@elasticmachine merge upstream

[kibanamachine]

Looks like this PR has a backport PR but it still hasn't been merged. Please merge it ASAP to keep the branches relatively in sync.

[bhavyarm]

@jloleysens do I need to start from 5.6.16 to test this PR? Thanks!

[jloleysens]

Hi @bhavyarm ! The PR description has a hacky version of to test the UI, but for an actual test we would need to load up a v7.7.0 stack (ES + Kibana) and have the following conditions met:

1. There is an index using known deprecated functionality that requires reindexing to fix
2. That same index is currently closed

The upgrade assistant should now be able to both detect the need for reindexing and that the index is closed. Reindexing should still be possible, though. We just want to let the user know that we need to open the index and re-close it.