Currently, if you try loading archive with index mappings not having
replica set into stateless ES, it won't work properly: you will get 503
error on calling `GET <index_name>/_stats`:
```
{
"error": {
"root_cause": [
{
"type": "no_shard_available_action_exception",
"reason": null
}
],
"type": "search_phase_execution_exception",
"reason": "all shards failed",
"phase": "query",
"grouped": true,
"failed_shards": [
{
"shard": 0,
"index": "indices-stats",
"node": null,
"reason": {
"type": "no_shard_available_action_exception",
"reason": null
}
}
]
},
"status": 503
}
```
In stateless replica is
[required](https://elastic.slack.com/archives/C037J0RKRAN/p1690218904855299)
in order to perform search requests (the "search shard").
This PR updates index mappings in es_archives with
`"auto_expand_replicas": "0-1"`, in order to be compatible with
stateless ES and so that we can re-use existing data sets rather than
creating new ones.
I checked with Core Team that we should fine to just adapt all mapping
files, but let me know if that doesn't work for you.
The same value is used to create the "real" SO
[indices](c79c09c3d0/packages/core/saved-objects/core-saved-objects-migration-server-internal/src/actions/constants.ts (L21))
in Kibana.
## Summary
Part of https://github.com/elastic/kibana/issues/161882
Updates all archives of the
`test/functional/fixtures/es_archiver/saved_objects_management` test
suite:
* Removing SO index definitions.
* Updating documents defined in `data.json`.
## Summary
Part of https://github.com/elastic/kibana/issues/161882
- Updates the remaining "dynamic" es archives (aka those using the
`$KIBANA_PACKAGE_VERSION` variable), deleting mappings.json.
- It also removes the piece of logic in `esArchiver` that replaces that
variable.
While searching in test files, I found few more places to replace
`PageObjects.timePicker.setAbsoluteRange` with
`PageObjects.common.setTime`.
Before changing I reviewed tests if it is not important to select
datePicker via UI component, but please double check scenarios because I
might miss the context.
## Summary
Part of https://github.com/elastic/kibana/issues/161882
Updates the `test/functional/fixtures/es_archiver/deprecations_service`
archive:
* deleting `mappings.json`
* updating documents in `data.json`
This way, _esArchiver_ will not delete SO indices and recreate them,
eliminating the odds of the related tests being flaky.
closes#113998
In order to make FTR functional tests faster and less flaky, we started
migrating time picker selection from UI action to Kibana API call:
```
await this.kibanaServer.uiSettings.update({
'timepicker:timeDefaults': `{ "from": <startTime>, "to": <endTime>}`,
});
```
In this PR I updated most of the tests listed in the meta issue, so we
can close it.
Flaky test runner
https://buildkite.com/elastic/kibana-flaky-test-suite-runner/builds/2606
## Summary
Previously the data plugin relied on the data view plugin to load saved
searches since the saved searches depend upon the data plugin and
circular dependencies needed to be avoided. This is innovative and
perhaps a bit crazy.
What this PR does
- Data view api no longer loads saved searches, removing browser saved
object client usage
- Moves `kibana_context` expression and getKibanaContext function from
data plugin to saved search plugin since it loads saved searches
- Rename data views `SavedObjectsClientCommon` to `PersistenceAPI` -
this is the abstraction around saved object loading that no longer is
exclusive to the saved objects api.
- Adds saved search server api (plugin contract) for loading saved
searches.
- Functional tests on browser and server for kibana_context expression
when loading saved searches
---------
Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com>
## Summary
Part of https://github.com/elastic/kibana/issues/158802
We decided that in the case we don't render a Lens suggestion (for
example when we have the SELECT * case) to render the histogram.
**Reminder**: Histogram makes sense only of there is a time field. For
text based mode, time field exists **ONLY** if there is the @timestamp
field.
I don't allow navigation to Lens or open the edit flyout in this case.
When the edit flyout allows the editing of the form based visualizations
(such as histogram) I will enable it then.
FTs have been changed to accomodate this change.
<img width="1751" alt="image"
src="3d28d881-bc60-43de-acf8-8cbcd172a3df">
### Checklist
- [x] [Unit or functional
tests](https://www.elastic.co/guide/en/kibana/master/development-tests.html)
were updated or added to match the most common scenarios
## Summary
Sometimes browser/driver process dies during test run on CI and FTR
fails with errors cascade, good example is
[here](https://buildkite.com/elastic/kibana-pull-request/builds/138535#0188fd74-9adf-4011-8168-1bdc6e3d0f17)
Current behaviour on `main`: FTR lifecycle hooks, defined in
[remote](57aea91fae/test/functional/services/remote/remote.ts)
service, has no access to the information about test suite run and
particular test failure. These hooks are related to WebDriver (browser)
state management and suppose to reset it to default state.
Currently we silently fail screenshot taking which means tests execution
is continued even if `--bail` flag is passed. It ends with cascade of
failures with the same error `NoSuchSessionError: invalid session id`
<details>
<summary>FTR output on failure</summary>
```
└- ✖ fail: discover/group1 discover test query should show correct time range string by timepicker
│ Error: expected 'Sep 19, 2015 @ 06:31:44.000' to equal 'Sep 23, 2015 @ 18:31:44.000'
│ at Assertion.assert (expect.js💯11)
│ at Assertion.apply (expect.js:227:8)
│ at Assertion.be (expect.js:69:22)
│ at Context.<anonymous> (_discover.ts:53:31)
│ at processTicksAndRejections (node:internal/process/task_queues:96:5)
│ at Object.apply (wrap_function.js:73:16)
│
│
└-> "after all" hook: afterTestSuite.trigger for "should reload the saved search with persisted query to show the initial hit count"
└- ✖ fail: discover/group1 discover test query "after all" hook: afterTestSuite.trigger for "should reload the saved search with persisted query to show the initial hit count"
│ NoSuchSessionError: invalid session id
│ at Object.throwDecodedError (node_modules/selenium-webdriver/lib/error.js:524:15)
│ at parseHttpResponse (node_modules/selenium-webdriver/lib/http.js:601:13)
│ at Executor.execute (node_modules/selenium-webdriver/lib/http.js:529:28)
│ at processTicksAndRejections (node:internal/process/task_queues:96:5)
│ at Task.exec (prevent_parallel_calls.ts:28:20)
│
│
└-> "after all" hook in "discover test"
│ debg Cleaning all saved objects { space: undefined }
│ succ deleted 2 objects
└-> "after all" hook: afterTestSuite.trigger in "discover test"
└- ✖ fail: discover/group1 discover test "after all" hook: afterTestSuite.trigger in "discover test"
│ NoSuchSessionError: invalid session id
│ at Object.throwDecodedError (node_modules/selenium-webdriver/lib/error.js:524:15)
│ at parseHttpResponse (node_modules/selenium-webdriver/lib/http.js:601:13)
│ at Executor.execute (node_modules/selenium-webdriver/lib/http.js:529:28)
│ at processTicksAndRejections (node:internal/process/task_queues:96:5)
│ at Task.exec (prevent_parallel_calls.ts:28:20)
│
│
└-> "after all" hook: unloadMakelogs in "discover/group1"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Unloading indices from "mappings.json"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Deleted existing index "logstash-2015.09.22"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Deleted existing index "logstash-2015.09.20"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Deleted existing index "logstash-2015.09.21"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Unloading indices from "data.json.gz"
└-> "after all" hook: afterTestSuite.trigger in "discover/group1"
└- ✖ fail: discover/group1 "after all" hook: afterTestSuite.trigger in "discover/group1"
│ NoSuchSessionError: invalid session id
│ at Object.throwDecodedError (node_modules/selenium-webdriver/lib/error.js:524:15)
│ at parseHttpResponse (node_modules/selenium-webdriver/lib/http.js:601:13)
│ at Executor.execute (node_modules/selenium-webdriver/lib/http.js:529:28)
│ at runMicrotasks (<anonymous>)
│ at processTicksAndRejections (node:internal/process/task_queues:96:5)
│ at Task.exec (prevent_parallel_calls.ts:28:20)
│
│
│
│0 passing (15.7s)
│4 failing
│
│1) discover/group1
│ discover test
│ query
│ should show correct time range string by timepicker:
│
│ Error: expected 'Sep 19, 2015 @ 06:31:44.000' to equal 'Sep 23, 2015 @ 18:31:44.000'
│ at Assertion.assert (expect.js💯11)
│ at Assertion.apply (expect.js:227:8)
│ at Assertion.be (expect.js:69:22)
│ at Context.<anonymous> (_discover.ts:53:31)
│ at processTicksAndRejections (node:internal/process/task_queues:96:5)
│ at Object.apply (wrap_function.js:73:16)
│
│
│2) discover/group1
│ discover test
│ query
│ "after all" hook: afterTestSuite.trigger for "should reload the saved search with persisted query to show the initial hit count":
│
│ NoSuchSessionError: invalid session id
│ at Object.throwDecodedError (node_modules/selenium-webdriver/lib/error.js:524:15)
│ at parseHttpResponse (node_modules/selenium-webdriver/lib/http.js:601:13)
│ at Executor.execute (node_modules/selenium-webdriver/lib/http.js:529:28)
│ at processTicksAndRejections (node:internal/process/task_queues:96:5)
│ at Task.exec (prevent_parallel_calls.ts:28:20)
│
│
│3) discover/group1
│ discover test
│ "after all" hook: afterTestSuite.trigger in "discover test":
│
│ NoSuchSessionError: invalid session id
│ at Object.throwDecodedError (node_modules/selenium-webdriver/lib/error.js:524:15)
│ at parseHttpResponse (node_modules/selenium-webdriver/lib/http.js:601:13)
│ at Executor.execute (node_modules/selenium-webdriver/lib/http.js:529:28)
│ at processTicksAndRejections (node:internal/process/task_queues:96:5)
│ at Task.exec (prevent_parallel_calls.ts:28:20)
│
│
│4) discover/group1
│ "after all" hook: afterTestSuite.trigger in "discover/group1":
│
│ NoSuchSessionError: invalid session id
│ at Object.throwDecodedError (node_modules/selenium-webdriver/lib/error.js:524:15)
│ at parseHttpResponse (node_modules/selenium-webdriver/lib/http.js:601:13)
│ at Executor.execute (node_modules/selenium-webdriver/lib/http.js:529:28)
│ at runMicrotasks (<anonymous>)
│ at processTicksAndRejections (node:internal/process/task_queues:96:5)
│ at Task.exec (prevent_parallel_calls.ts:28:20)
```
</details>
This PR change: I didn't find a good reason why we need to fail silently
on screenshot taking. I added a check WebDriver session status with
`hasOpenWindow` and take failure artefacts only if is still valid.
Next change is to fail FTR after hooks related to WebDriver silently:
there is no help having cascade of the repeated stacktrace so I wrap
WebDriver call in hooks with `tryWebDriverCall` that catches the error
and only prints it for visibility.
<details>
<summary>FTR new output on failure</summary>
```
│ERROR WebDriver session is no longer valid.
│ Probably Chrome process crashed when it tried to use more memory than what was available.
│ERROR Browser is closed, no artifacts were captured for the failure
└- ✖ fail: discover/group1 discover test query should show correct time range string by timepicker
│ Error: expected 'Sep 19, 2015 @ 06:31:44.000' to equal 'Sep 23, 2015 @ 18:31:44.000'
│ at Assertion.assert (expect.js💯11)
│ at Assertion.apply (expect.js:227:8)
│ at Assertion.be (expect.js:69:22)
│ at Context.<anonymous> (_discover.ts:53:31)
│ at processTicksAndRejections (node:internal/process/task_queues:96:5)
│ at Object.apply (wrap_function.js:73:16)
│
│
└-> "after all" hook: afterTestSuite.trigger for "should reload the saved search with persisted query to show the initial hit count"
│ERROR WebDriver session is no longer valid
└-> "after all" hook in "discover test"
│ debg Cleaning all saved objects { space: undefined }
│ warn browser[SEVERE] ERROR FETCHING BROWSR LOGS: This driver instance does not have a valid session ID (did you call WebDriver.quit()?) and may no longer be used.
│ succ deleted 2 objects
└-> "after all" hook: afterTestSuite.trigger in "discover test"
│ERROR WebDriver session is no longer valid
└-> "after all" hook: unloadMakelogs in "discover/group1"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Unloading indices from "mappings.json"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Deleted existing index "logstash-2015.09.22"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Deleted existing index "logstash-2015.09.20"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Deleted existing index "logstash-2015.09.21"
│ info [test/functional/fixtures/es_archiver/logstash_functional] Unloading indices from "data.json.gz"
└-> "after all" hook: afterTestSuite.trigger in "discover/group1"
│ERROR WebDriver session is no longer valid
0 passing (16.2s)
1 failing
1) discover/group1
discover test
query
should show correct time range string by timepicker:
Error: expected 'Sep 19, 2015 @ 06:31:44.000' to equal 'Sep 23, 2015 @ 18:31:44.000'
at Assertion.assert (expect.js💯11)
at Assertion.apply (expect.js:227:8)
at Assertion.be (expect.js:69:22)
at Context.<anonymous> (_discover.ts:53:31)
at processTicksAndRejections (node:internal/process/task_queues:96:5)
at Object.apply (wrap_function.js:73:16)
```
</details>
Flaky-test-runner verification: started 100x to hopefully catch invalid
session on CI
https://buildkite.com/elastic/kibana-flaky-test-suite-runner/builds/2539
Note: locally I was simulating it by calling `this.driver.close()` to
close browser before screenshot taking
## Summary
Closes#154500.
Closes https://github.com/elastic/kibana/issues/114206.
This updates the UX around cloning dashboards. I removed the clone
confirm model, so cloning a dashboard is just a one click action now.
This aligns with how other apps like ML handle cloning. I've also made
the dashboard title breadcrumb a button in edit mode that opens the
dashboard settings flyout for better discoverability.
4f5ea117-a5e4-4ec5-9113-8b09fd8c84a1
I also changed the pattern for cloned dashboard title from `Dashboard
Title Copy` to `Dashboard Title (#)`.
<img width="1226" alt="Screenshot 2023-06-30 at 1 03 35 PM"
src="b50ba5c6-dc95-4aab-a320-b1a78b74c1b6">
## Summary
Previously - If you changed a data view's index pattern AND the new
pattern didn't contain the timestamp field, you'd see a blank timestamp
field and it would let you save. The data view would have been saved
with the previous timestamp field which doesn't exist.
Now - The timestamp validator checks to make sure the selected timestamp
field is in the list of available options. This is helpful because it
keeps the previous timestamp value in case you do select an index
pattern that contains it.
Closes: https://github.com/elastic/kibana/issues/150219
- Closes https://github.com/elastic/kibana/issues/150317
## Summary
This PR:
- hides "Use without saving" button from NoData screen on Discover page
to align with Dashboard and Lens behaviour
- allows to open a saved search via URL even if there are no data views
present (before NoData screen was shown instead).
## Summary
Composite runtime fields with dots in the name were broken, now fixed.
To verify -
1. Create runtime field with dot in the name
2. Make a composite runtime field with subfields with dot in the name
3. Go back and edit those fields
Closes: https://github.com/elastic/kibana/issues/159648
### Release note
Fixes creation and editing of composite runtime fields with dots in the
names.
## Summary
While splitting `test/functional/apps/dashboard/group2/config.ts` in
#160612, I noticed that some tests (`it` function level) takes 30-45
seconds though having little actions to do in Kibana.
After closer look I found that:
- `await PageObjects.dashboard.clickNewDashboard();` call takes ~13-15
seconds when dashboard is empty
- `await dashboardAddPanel.closeAddPanel();` call takes ~12 seconds when
there are no flyouts displayed
Expected improvement:
- `await PageObjects.dashboard.clickNewDashboard()` call takes **2**
seconds, used **103** times in tests.
- `await dashboardAddPanel.closeAddPanel()` call takes **3** seconds,
used **21** times in tests.
PR FTR configs runtime
<img width="778" alt="image"
src="673fee19-91ef-4bc3-9848-5f844b42774c">
`main` branch last 3 days avg configs runtime
<img width="1555" alt="image"
src="6e364cb9-c786-411a-9491-a749351f4c94">
flaky test runner 50x for 6 most affected configs
https://buildkite.com/elastic/kibana-flaky-test-suite-runner/builds/2519
## Summary
Splitting `test/functional/apps/dashboard/group2/config.ts` as it
getting close to 35 minutes and quite often run on its own on CI worker.
<img width="1713" alt="image"
src="ac7f5dc6-2a12-4057-af98-81ff53bac1c4">`
This PR splits config into 2 almost run time equal groups:
- test/functional/apps/dashboard/group2/config.ts 18m 31s
- test/functional/apps/dashboard/group6/config.ts 16m 53s
Flaky-test-runner for both configs:
https://buildkite.com/elastic/kibana-flaky-test-suite-runner/builds/2522
Optimize switch to text-based queries by reducing unnecessary EuiDataGrid renderings and addressing performance regression. Introduce a PARTIAL state emission in the document$ observable for text-based query results. This state indicates ongoing loading in the UI. The appState to URL changes hook completes with the COMPLETE state submission to document$, enabling UI rendering.
Co-authored-by: Davis McPhee <davismcphee@hotmail.com>
- Closes https://github.com/elastic/kibana/issues/160212
## Summary
Added the missing encoding and updated tests.
## For testing
An id with special characters like `/` can be specified via `_bulk` API.
Example:
```
POST _bulk
{ "index" : { "_index" : "context-special", "_id" : "test&?#+/=" } }
{ "timestamp":"2023-06-20T20:00:00.123Z", "name": "test" }
```
Closes https://github.com/elastic/kibana/issues/159353
Closes https://github.com/elastic/kibana/issues/159756
## Summary
This PR adds styling so that, when the dashboard's chrome is hidden or
when there is a header banner present, the dashboard's top navigation
bar's `top` position is adjusted as necessary - this prevents it from
either (a) getting hidden behind the Kibana chrome or (b) floating in
the wrong position and overlapping the dashboard content, regardless of
if the dashboard is in embed mode (i.e. `embed=true` is present in the
URL) or not (i.e. `embed=false` is present in the URL **or**, more
commonly, there is no `embed` parameter in the URL).
### Embed Mode
- #### Before:
76d3c70c-936c-4bcc-985c-4fb433f0cff3
- #### After:
137bc103-666b-4fdd-ab4e-8994345e21b4
It also resolves a bug where the `isEmbeddedExternally` component state
was never actually being set, which meant that the
`ExitFullScreenButton` was always receiving `toggleChrome=true`. This
made it so that entering and exiting fullscreen mode in an embedded
dashboard would force the chrome to be visible (which should never
happen in embed mode).
#### How to Test
The easiest way to test this PR is to simply add `embed=true` to your
dashboard URL - because this PR also fixes
https://github.com/elastic/kibana/issues/159756, this will essentially
mimic the embedded experience.
Alternatively, if you want to test this in an actual iframe...
1. Start and login to Kibana with the default `kibana.yml` settings
2. Create and embed a dashboard using an iframe in an HTML file and open
that file in your browser - the iframe will show a prompt to login, but
you won't be able to. Instead...
3. Add the following settings to your `kibana.yml` file:<br><Br>
```
xpack.security.secureCookies: true
xpack.security.sameSiteCookies: 'None'
```
4. Wait for Kibana to re-load
5. Refresh the HTML page from step 2
6. The embedded dashboard should now be available for you to test 👍
#### Scenarios Tested
- **Non-fullscreen mode**
<details>
<summary>✅ Without filter pill, without header banner <i>(click to see
screenshot)</i></summary>
<img
src="f68bbcfb-74d8-497c-a2ae-33e8e0c02660"/>
</details>
<details>
<summary>✅ Without filter pill, with header banner <i>(click to see
screenshot)</i></summary>
<img
src="7c19711c-61dc-499a-b1d0-01fab639a27e"/>
</details>
<details>
<summary>✅ With filter pill, without header banner <i>(click to see
screenshot)</i></summary>
<img
src="36e848bd-f0d9-41e3-8a8a-a48571ad5cd2"/>
</details>
<details>
<summary>✅ With filter pill, with header banner <i>(click to see
screenshot)</i></summary>
<img
src="cd7489f6-3f34-439a-a30e-3ef39f3970b5"/>
</details>
<details>
<summary>✅ With filter pill, with header banner <b>and</b> notification
banner <i>(click to see GIF)</i></summary>
<img
src="bd67b4eb-4f68-4d9b-9e22-4d1b2d2e4d90"/>
</details>
- **Fullscreen mode**
<details>
<summary>✅ Without filter pill, without banner <i>(click to see
screenshot)</i></summary>
<img
src="d7d15560-7698-424f-b761-59b5557abe37"/>
</details>
<details>
<summary>✅ Without filter pill, with header banner <i>(click to see
screenshot)</i></summary>
<img
src="311b6f3d-5152-4d16-ba39-160978c60c96"/>
</details>
<details>
<summary>✅ With filter pill, without header banner <i>(click to see
screenshot)</i></summary>
<img
src="bff9e040-8169-40c7-a086-13a19e870383"/>
</details>
<details>
<summary>✅ With filter pill, with header banner <i>(click to see
screenshot)</i></summary>
<img
src="3f453811-e65d-4ac4-9524-c396f9efdbdd"/>
</details>
<details>
<summary>✅ With filter pill, with header banner <b>and</b> notification
banner <i>(click to see GIF)</i></summary>
<img
src="f79673e7-03f2-49fa-be56-b67bf7a12976"/>
</details>
### Non-Embed Mode
- #### Before:
71ffc964-2844-41a6-98d6-353e84d674be
- #### After:
894aa292-b611-4e5e-a0d7-fe3d256fc3ba
### Checklist
- [x] [Unit or functional
tests](https://www.elastic.co/guide/en/kibana/master/development-tests.html)
were updated or added to match the most common scenarios
- [x] This was checked for [cross-browser
compatibility](https://www.elastic.co/support/matrix#matrix_browsers)
### For maintainers
- [ ] This was checked for breaking API changes and was [labeled
appropriately](https://www.elastic.co/guide/en/kibana/master/contributing.html#kibana-release-notes-process)
Closes https://github.com/elastic/kibana/issues/160057
## Summary
It **seems** like the failure was a one-off or, at the very least, very
rare - I [ran the test through the flaky test
runner](https://buildkite.com/elastic/kibana-flaky-test-suite-runner/builds/2423)
without any changes and it passed 100 times. That being said, I was able
to replicate the failure locally by switching to the preset "Slow 3G"
network before running the test - this means, if the CI environment is
running especially slow for any given reason, this test has the
potential to fail again.
So, to be safe, I added an extra `await` for the Kibana **chrome** to
show up after every hard refresh - previously we were only waiting for
the header loading state to go away, which doesn't work when Kibana
itself is taking a little longer than expected to load and the Kibana
chrome (where the header loading state is located) isn't yet visible.
This **also** [passed the flaky test runner 100
times](https://buildkite.com/elastic/kibana-flaky-test-suite-runner/builds/2425)
but, as a bonus, the test **passed locally** when throttling the network
using the same "Slow 3G" preset as before - hopefully this means that
the test is less likely to fail even when the CI environment is running
slow 👍
### Checklist
- [x] [Unit or functional
tests](https://www.elastic.co/guide/en/kibana/master/development-tests.html)
were updated or added to match the most common scenarios
### For maintainers
- [ ] This was checked for breaking API changes and was [labeled
appropriately](https://www.elastic.co/guide/en/kibana/master/contributing.html#kibana-release-notes-process)
