[QA] Switch tests to use importExport - discover

[wayneseymour]

Switch es archiver for kbn archiver everywhere possible

Partially closes #93797

[wayneseymour]

@elasticmachine merge upstream

[wayneseymour]

@spalger hey sir, Lee and I think we could use your expertise. CI group 6 fails consistently on ci, but never on my local. I wrote a fn that does much the same and it worked. But, we dont wanna use that, we wanna use the kbnarchiver. I'll set it up the way it was and hopefully you can free us of this pestilence! hahaaa

[spalger]

The error I'm seeing in ciGroup6 is because there is no dirFile export in test/functional/utils/import_data.ts, which seems to be the replacement you wrote. Can you point me to the code using kbnArchiver so I can take a look at what's going on?

[wayneseymour]

The error I'm seeing in ciGroup6 is because there is no dirFile export in test/functional/utils/import_data.ts, which seems to be the replacement you wrote. Can you point me to the code using kbnArchiver so I can take a look at what's going on?

yeah totally, I've got another draft pr that has less of my goofs within: #95667

i need it to go through ci once more so you can see that I was talking about though. Running now.

[wayneseymour]

Merged in from: https://github.com/elastic/kibana/pull/95667/files

[wayneseymour]

Since I've to throw an error in lieu of how to use the logger properly, I'm going to record the output of the saved object "types" within the .kibana index here.
Since we've at least three separate calls, so we'll have at least that many ci runs as we can only throw the error once per run.

1st Run BEFORE await esArchiver.load('empty_kibana');

Error: 
[00:00:04]               │      Error: 
[00:00:04]               │ ### SAVED OBJECT TYPES IN index: [.kibana]: 
[00:00:04]               │ [
[00:00:04]               │   {
[00:00:04]               │     doc_count: 1,
[00:00:04]               │     key: 'config'
[00:00:04]               │   }
[00:00:04]               │ ]

2nd Run AFTER await esArchiver.load('empty_kibana'); AND BEFORE await kibanaServer.importExport.load('discover')

[00:00:04]               │      Error: 
[00:00:04]               │ ### SAVED OBJECT TYPES IN index: [.kibana]: 
[00:00:04]               │ [
[00:00:04]               │   {
[00:00:04]               │     doc_count: 2,
[00:00:04]               │     key: 'config'
[00:00:04]               │   }
[00:00:04]               │ ]

3rd Run AFTER await kibanaServer.importExport.load('discover')

Error: 
### SAVED OBJECT TYPES IN index: [.kibana]: 
[
  {
    doc_count: 2,
    key: 'config'
  },
  {
    doc_count: 1,
    key: 'core-usage-stats'
  },
  {
    doc_count: 1,
    key: 'index-pattern'
  },
  {
    doc_count: 1,
    key: 'search'
  }
]

@LeeDr what do you think sir?

[LeeDr]

That output is in the debug logs when a test fails so I don't think you have to force the test to fail at each step. We're adding this logging of the saved objects to help debug why the test was failing in CI. The thought was that a previous test was unloading the .kibana index and that's why the saved object import was failing.

So maybe the calls to log the saved object aggregation could happen in the kibanaServer.importExport.load() method? But I'm not sure if that adds seconds to the call when we're trying to optimize the test execution time?

[LeeDr]

I'm trying to understand the error reason: 'no such index [.kibana_8.0.0] and [require_alias] request flag is [true] and [.kibana_8.0.0] is not an alias',

I think the importExport.load should be loading into .kibana and in that case [require_alias] would be OK. Is it trying to load into .kibana_8.0.0 ?

    before(async function () {
      await kibanaServer.savedObjects.clean({ types: ['search', 'index-pattern'] });
      await kibanaServer.importExport.load('discover');

[00:08:59]               │ proc [kibana]   log   [20:25:08.064] [info][repository][savedobjects-service] savedObjects.bulkCreate success {
[00:08:59]               │ proc [kibana]   saved_objects: [
[00:08:59]               │ proc [kibana]     {
[00:08:59]               │ proc [kibana]       id: 'logstash-*',
[00:08:59]               │ proc [kibana]       type: 'index-pattern',
[00:08:59]               │ proc [kibana]       error: {
[00:08:59]               │ proc [kibana]         statusCode: 500,
[00:08:59]               │ proc [kibana]         error: 'Internal Server Error',
[00:08:59]               │ proc [kibana]         message: 'An internal server error occurred'
[00:08:59]               │ proc [kibana]       },
[00:08:59]               │ proc [kibana]       rawResponse: {
[00:08:59]               │ proc [kibana]         error: {
[00:08:59]               │ proc [kibana]           type: 'index_not_found_exception',
[00:08:59]               │ proc [kibana]           reason: 'no such index [.kibana_8.0.0] and [require_alias] request flag is [true] and [.kibana_8.0.0] is not an alias',
[00:08:59]               │ proc [kibana]           index_uuid: '_na_',
[00:08:59]               │ proc [kibana]           index: '.kibana_8.0.0'
[00:08:59]               │ proc [kibana]         },
[00:08:59]               │ proc [kibana]         _index: '.kibana_8.0.0',
[00:08:59]               │ proc [kibana]         _id: 'index-pattern:logstash-*',
[00:08:59]               │ proc [kibana]         status: 404
[00:08:59]               │ proc [kibana]       }
[00:08:59]               │ proc [kibana]     },

[LeeDr]

Making progress!

From the discover/index file, these passed;

    loadTestFile(require.resolve('./_saved_queries'));
    loadTestFile(require.resolve('./_discover'));
    loadTestFile(require.resolve('./_discover_histogram'));
    loadTestFile(require.resolve('./_doc_table'));
    loadTestFile(require.resolve('./_filter_editor'));
    loadTestFile(require.resolve('./_errors'));
    loadTestFile(require.resolve('./_field_data'));
    loadTestFile(require.resolve('./_field_data_with_fields_api'));
    loadTestFile(require.resolve('./_shared_links'));
    loadTestFile(require.resolve('./_sidebar'));
    loadTestFile(require.resolve('./_source_filters'));
    loadTestFile(require.resolve('./_large_string'));

Failed here on _inspector
    loadTestFile(require.resolve('./_inspector'));
    loadTestFile(require.resolve('./_doc_navigation'));
    loadTestFile(require.resolve('./_date_nanos'));
    loadTestFile(require.resolve('./_date_nanos_mixed'));
    loadTestFile(require.resolve('./_indexpattern_without_timefield'));
    loadTestFile(require.resolve('./_discover_fields_api'));
    loadTestFile(require.resolve('./_data_grid'));
    loadTestFile(require.resolve('./_data_grid_context'));
    loadTestFile(require.resolve('./_data_grid_field_data'));
    loadTestFile(require.resolve('./_data_grid_doc_navigation'));
    loadTestFile(require.resolve('./_data_grid_doc_table'));
    loadTestFile(require.resolve('./_indexpattern_with_unmapped_fields'));
    loadTestFile(require.resolve('./_runtime_fields_editor'));
    loadTestFile(require.resolve('./_huge_fields'));

[spalger]

This should be in the suite that hasn't been migrated yet, rather than in the suite which has been migrated to use importExport and not replace the .kibana index at runtime.

[wayneseymour]

@spalger the suite before this one is skipped bud. Which makes this the first test run. Whatcha think?

[spalger]

If this is the first suite run then the .kibana index already exists and is empty and this is unnecessary.

[wayneseymour]

Ok, I'll strip it and see what happens.

[kibanamachine]

💚 Build Succeeded

• continuous-integration/kibana-ci/pull-request
• Commit: 6ee3352
• Storybooks Preview
• Documentation Changes

Metrics [docs]

✅ unchanged

History

• 💚 Build #121206 succeeded 32bc174
• 💔 Build #121100 failed ee42d35
• 💔 Build #121093 failed 7eabe9a
• 💚 Build #120147 succeeded 70299a6
• 💚 Build #120100 succeeded e94f467

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

cc @wayneseymour

[spalger]

LGTM

[LeeDr]

LGTM - Code reviewed and Jenkins results