text
stringlengths 20
57.3k
| labels
class label 4
classes |
|---|---|
Title: add ExecWithOptions to exec_util.go for ClientSet Creation
Body: We should fine grained debug statements to ExecWithOptions in the framework exec_util.go class , so that in tests where theres alot of execs where hangs can happen i.e. on windows, we can see where tests are getting clogged up. Mostly i want this for network_policy.go, but i suppose its a generic improvement
```
Logf("ExecWithOptions: Clientset creation ")
``` | 0easy
|
Title: Improve concurrency for hack/verify-structured-logging.sh
Body: <!-- Feature requests are unlikely to make progress as an issue.
Instead, please suggest enhancements by engaging with SIGs on slack and mailing lists.
A proposal that works through the design along with the implications of the change can be opened as a KEP:
https://git.k8s.io/enhancements/keps#kubernetes-enhancement-proposals-keps
-->
#### What would you like to be added:
Improve concurrency for this script https://github.com/kubernetes/kubernetes/blob/master/hack/verify-structured-logging.sh#L52
follow up from this PR: https://github.com/kubernetes/kubernetes/pull/99090
#### Why is this needed:
Speed up the execution time.
/sig instrumentation testing | 0easy
|
Title: [google_sign_in_web] Move js-interop layer out of `generated`
Body: The js-interop layer of the web version of Google Sign In used to be generated by the [js_facade_gen script](https://github.com/dart-lang/js_facade_gen), but it is no longer.
## Proposal
Move the js-interop layer ([`gapi.dart`, `gapiauth2.dart`...](https://github.com/flutter/plugins/tree/main/packages/google_sign_in/google_sign_in_web/lib/src/generated)) outside of `generated` (or rename `generated` to `js_interop` or similar), so the files don't receive any special treatment.
Also: fix tests, and whatever issues the analyzer may start having with the change, of course. | 0easy
|
Title: Consider adding checks to make sure the "filtering" layer's children is not empty.
Body: Consider adding checks to make sure the "filtering" layer's children is not empty. just like what we did in opacity layer.
* ImageFilterLayer
* ColorFilterLayer
* ShaderMaskLayer
c.f.
https://github.com/flutter/engine/blob/e444009bf46b39af0b2a2aa6e7c28c668c50004c/flow/layers/opacity_layer.cc#L40
| 0easy
|
Title: `ignore_raster_cache` on `CompositorContext::ScopedFrame::Raster` seems unused.
Body: This seems like a defunct API that can be removed to cleanup the code in the rasterizer. | 0easy
|
Title: TileMode images missing in API docs
Body: https://master-api.flutter.dev/flutter/dart-ui/TileMode-class.html is missing sweep gradient images
Everywhere else that shows these images (dart:ui Gradient constructors, painting lib Gradient subclass tileMode properties) omit the clamp version of TileMode. | 0easy
|
Title: Add //flutter/fml/math.h with common math constants.
Body: This came up recently with the use of `cmath` on Windows requiring `_USE_MATH_DEFINES` to be defined before common constants could be accessed.
This define either needs to define at the toolchain level or wired up in the private configs for the targets that use it or the public configs of the target that specify it in their headers.
This platform specific oddity build oddity is hard for developers to reason about. We just just define an use our own constants. Something like would work well. Additional math utilities may be added here.
```
// e
constexpr double kE = 2.7182818284590452354;
// log_2 e
constexpr double kLog2_E = 1.4426950408889634074;
// log_10 e
constexpr double kLog10_E = 0.43429448190325182765;
// log_e 2
constexpr double klogE_2 = 0.69314718055994530942;
// log_e 10
constexpr double klogE_10 = 2.30258509299404568402;
// pi
constexpr double kPi = 3.14159265358979323846;
// pi/2
constexpr double kPiOver2 = 1.57079632679489661923;
// pi/4
constexpr double kPiOver4 = 0.78539816339744830962;
// 1/pi
constexpr double k1OverPi = 0.31830988618379067154;
// 2/pi
constexpr double k2OverPi = 0.63661977236758134308;
// 2/sqrt(pi)
constexpr double k2OverSqrtPi = 1.12837916709551257390;
// sqrt(2)
constexpr double kSqrt2 = 1.41421356237309504880;
// 1/sqrt(2)
constexpr double k1OverSqrt2 = 0.70710678118654752440;
``` | 0easy
|
Title: Consider removing saveCompilationTrace
Body: `dart:ui` exposes [saveCompilationTrace](https://api.flutter.dev/flutter/dart-ui/saveCompilationTrace.html), which, according to @jonahwilliams, is:
> completely unused (and unusable) by the framework so we should probably file a bug to remove it
Source: https://github.com/flutter/engine/pull/18933#discussion_r438408808
| 0easy
|
Title: [Reporting] Clean up Report Info Flyout file to use imported strings
Body: The `report_info_flyout_content` has breached a point where it is more convenient for maintainers of the file to source the user-facing content strings from a separate file.
This issue is to follow the imports added in https://github.com/elastic/kibana/pull/130020/files#diff-83b67268eba04e092077b198cf34a3a7c273878024670feac96ca495fc8d829f to hoist all the user-facing content strings from the external file. | 0easy
|
Title: [Usage collection] Remove deprecated rollups
Body: Follow up #127744.
In 8.0.0, we should have safely deleted the following rollup logic in the `kibana_usage_collection`:
- [ ] Application Usage's transactional to daily rollups: https://github.com/elastic/kibana/blob/4681a80317176543535bfdb2188d94aa40d741fc/src/plugins/kibana_usage_collection/server/collectors/application_usage/rollups/daily.ts
- [ ] UI Metrics to UI Counters rollup: https://github.com/elastic/kibana/blob/577ccfb6f8fbb56e9c2d8de4daccc0698773fa6f/src/plugins/kibana_usage_collection/server/collectors/ui_counters/rollups/rollups.ts#L40-L90 | 0easy
|
Title: [Search][Expressions] Remove `AbortController` polyfill
Body: Kibana's node version is at 16 now so we can remove `AbortController` polyfills:
https://github.com/elastic/kibana/blob/66aca8b486342727de3f5c95821bbeb22fae3d29/src/plugins/data/server/lib/get_request_aborted_signal.ts#L11
https://github.com/elastic/kibana/blob/66aca8b486342727de3f5c95821bbeb22fae3d29/src/plugins/expressions/common/execution/execution.ts#L79 | 0easy
|
Title: [Task Manager] Clean up performance tests
Body: With https://github.com/elastic/kibana/pull/117294, we are removing the `perf_hooks` in task manager due to possible memory leakage. There is also a set of skipped functional tests in https://github.com/elastic/kibana/blob/main/x-pack/test/plugin_api_perf/test_suites/task_manager/index.ts that we should make sure to remove once we've confirmed that 🤞 removing the perf_hooks fixes the memory issue. | 0easy
|
Title: [Alerting] Update internal API calls use new Alerting APIs
Body: There are still places in the code that call the legacy alerting APIs. Now that we are using telemetry to track usages of the legacy APIs, we should update to use the new ones so that the telemetry numbers are not thrown off.
https://github.com/elastic/kibana/blob/46bfe577c5823d759820141e0665e27bc1a12792/x-pack/examples/alerting_example/public/components/view_alert.tsx#L39-L46
https://github.com/elastic/kibana/blob/46bfe577c5823d759820141e0665e27bc1a12792/x-pack/examples/alerting_example/public/components/view_astros_alert.tsx#L45-L52
https://github.com/elastic/kibana/blob/46bfe577c5823d759820141e0665e27bc1a12792/x-pack/plugins/alerting/public/alert_api.ts#L16-L37
| 0easy
|
Title: [actions] rename ServiceNow `isLegacy` property to `usesTableApi` (or similar)
Body: Mike Cote and I discussed some of the aspects of the `isLegacy` config property that got added to the ServiceNow connector in [PR 105440](https://github.com/elastic/kibana/pull/105440). One of them is that `isLegacy` doesn't "scale well" over time, if we end up having to make some other change like this later, and have to come up with a `isLegacy2` field as well. It's also not very descriptive of what it actually means.
We think rather than use a generic term like "legacy", we should probably be in the mode of describing the scenario more specifically instead. For instance, we could rename this field `usesTableApi` (the old API), with a default of `false` and the migration changing the old connectors to set this value to `true`. No changes to the value, just a change to the name. This makes it both "scale better" for other such properties we might need later, as well as being more descriptive in the code, as well as anyone looking at the raw connector data.
There don't seem to be a lot of references to the property in the `actions` and `triggers_actions_ui` plugins, so seems like something that would be straight-forward to fix.
This feels kinda low priority to me, and also that if we don't do it before 7.16, we probably don't want to do it at all, because it would then require ANOTHER migration, which seems like overkill. Perhaps it's just a "lesson learned" about a pattern we should avoid in the future ... | 0easy
|
Title: Rename functions in aggregated_transactions helper
Body: Some functions in the helper include `ForAggregatedTransactions` in the name but they are meant to be used for both single transactions and aggregated transactions.
The term `aggregated` can be removed from such methods. (e.g. https://github.com/elastic/kibana/blob/a2ac439f56313b7a3fc4708f54a4deebf2615136/x-pack/plugins/apm/server/lib/helpers/aggregated_transactions/index.ts#L81).
In addition, they could be moved to another helper. | 0easy
|
Title: [APM] Remove unused transaction breakdown aggregation
Body: It looks like the following transaction breakdown aggregation is unused:
https://github.com/elastic/kibana/blob/00bb59713ed115343eb70d4e39059476edafbade/x-pack/plugins/apm/server/lib/transactions/breakdown/index.ts#L54-L58
TODO:
- [ ] Verify whether it's used anywhere. If not remove it.
- [ ] Rename the function `getTransactionBreakdown` to `getSpanBreakdown` | 0easy
|
Title: [APM] Prefix internal APM APIs with `/internal/`
Body: All APM apis with very few exceptions should be marked as internal. We can do that easily by prefxing a route with `/internal/`. APM apis are defined in: [`x-pack/plugins/apm/server/routes`](https://github.com/elastic/kibana/tree/9dff4d0827e8e1424ca4d86f692dec6c738f84c3/x-pack/plugins/apm/server/routes)
Public APIs:
- [Agent configuration APIs](https://www.elastic.co/guide/en/kibana/master/agent-config-api.html)
- [Annotations APIs](https://www.elastic.co/guide/en/kibana/master/apm-annotation-api.html)
- [Source maps](https://www.elastic.co/guide/en/kibana/master/rum-sourcemap-api.html)
Additional context: https://docs.google.com/document/d/1v4R9hWiQObP8sERfflGScC9Yp36fLE0wVlUNf13mVrc/edit#heading=h.2b1v1tr0ep8m | 0easy
|
Title: [Alerting] Jest test suites with unhandled promise rejections
Body: See https://github.com/elastic/kibana/issues/112699 | 0easy
|
Title: [APM] Improve function name: getIsUsingTransactionEvents
Body: In order to show the callout, the notifies the user when they are using transaction data instead of metrics we create a function called [getIsUsingTransactionEvents](https://github.com/elastic/kibana/blob/master/x-pack/plugins/apm/server/lib/helpers/aggregated_transactions/get_is_using_transaction_events.ts#L15). But this name doesn't match what this function returns, e.g. when `SearchAggregatedTransactionSetting is never` all data is queried from the transaction index, but this function returns false.
We could call it `should_display_fallback_message` or open to suggestions.
Also, there's an opportunity to combine some of the logic from [`getSearchAggregatedTransactions` ](https://github.com/elastic/kibana/blob/master/x-pack/plugins/apm/server/lib/helpers/aggregated_transactions/index.ts#L53): While `getIsUsingTransactionEvents` depends on `getSearchAggregatedTransactions` while both are making similar checks on the `xpack.apm.searchAggregatedTransactions` config value. In code path, calling `getIsUsingTransactionEvents` may call `getSearchAggregatedTransactions` which runs an ES query, then calls `getHasTransactions` which runs another ES query. We should avoid this if possible. | 0easy
|
Title: [alerting] wording in execution context properties contains the legacy term "Alert"
Body: Relates to https://github.com/elastic/kibana/issues/90375.
In PR https://github.com/elastic/kibana/pull/108197, support was added to alerting and task manager to set the "executionContext", so that calls to ES will be sent with a header indicating the source of the ES call (task, alerting rule, etc).
Realized after the merge, that "alert" leaked through 🤦🏻.
I also thought we might want to tweak the strings that are being used anyway. We should check how they are appearing in ES logs I guess, make sure they make sense.
Additionally - could be split into a separate PR - connectors have not be instrumented like this, but should be. Presumably they would only augment the index connector, so I'm guessing these would be easy to identify from ES logs anyway. So I don't think there's a lot of urgency for that.
Here are where the strings are currently set:
https://github.com/elastic/kibana/blob/36bba6ffe0aa4ab532ed857946205ee98ba2d5a4/x-pack/plugins/task_manager/server/task_running/task_runner.ts#L272-L277
https://github.com/elastic/kibana/blob/36bba6ffe0aa4ab532ed857946205ee98ba2d5a4/x-pack/plugins/task_manager/server/task_running/ephemeral_task_runner.ts#L200-L205
https://github.com/elastic/kibana/blob/36bba6ffe0aa4ab532ed857946205ee98ba2d5a4/x-pack/plugins/alerting/server/task_runner/task_runner.ts#L265-L272 | 0easy
|
Title: [Task Manager] [8.0] Add migrations for "actions:* and "alerting:*" task types
Body: Relates to https://github.com/elastic/kibana/issues/100067
Starting in 8.0, we need to ensure that the task manager task documents are updated to point to the new, migrated saved object IDs for actions and alerting task types. This is only applicable for saved objects that live in a custom space - saved objects in the default space are not affected. Both task types contain a `spaceId` param that will tell us this.
Alerting task types contain an `alertId` params that needs to be migrated from a legacy ID to a migrated saved object ID using [this newly added API](https://github.com/elastic/kibana/pull/107767).
Action task types contain an `actionTaskParamsId ` params that needs to be migrated from a legacy ID to a migrated saved object ID using [this newly added API](https://github.com/elastic/kibana/pull/107767).
See [the PoC PR](https://github.com/elastic/kibana/pull/107611) for more insight into how this might work. | 0easy
|
Title: [Event log][7.x] Update event log client to search across legacy IDs
Body: Relates to https://github.com/elastic/kibana/issues/100067
Blocked by https://github.com/elastic/kibana/issues/56771
In https://github.com/elastic/kibana/issues/108075, we are adding a new field to every rule saved object that points to the legacy ID (which is the current ID in 7.x but will potentially differ from the ID in 8.0). To support this new field, we need to change the event log client to accept this new ID field and use it in the query to ensure we find all relevant logs.
There is a chance that we will encounter a conflict scenario, where a legacy ID matches some migrated ID. In this case, we need to rely on [this change which will add the version to the event log](https://github.com/elastic/kibana/issues/56771) and handle that in our query appropriately.
See [the PoC PR](https://github.com/elastic/kibana/pull/107611) for more insight into how this might work. | 0easy
|
Title: [Alerting][7.x] Add migration that copies existing rule saved object ids to another location in the saved object
Body: Relates to https://github.com/elastic/kibana/issues/100067
To prepare for the upcoming saved object ID migration, we need to copy over the existing ID to a new location in the saved object for every rule.
Once the migration happens in 8.0, there is no way to determine the legacy ID given the new, migrated ID. This is a problem for [our code that searches against the event log](https://github.com/elastic/kibana/blob/master/x-pack/plugins/alerting/server/rules_client/rules_client.ts#L455) as the event log will contain references to legacy IDs and we need to start searching across both IDs.
See [the PoC PR](https://github.com/elastic/kibana/pull/107611) for more insight into how this might work. | 0easy
|
Title: [Alerting][7.x] Update rules detail page to use new saved object resolve API when necessary
Body: Relates to https://github.com/elastic/kibana/issues/100067
When we load a rule on the rule detail page, we need to ensure we are properly calling the new `savedObjectsClient.resolve()` API which will return not only the saved object, but also information about how the saved object ID resolved due to [the migration happening across all saved objects](https://github.com/elastic/kibana/issues/100067). This will ensure that users loading the page with an outdated ID will still see the appropriate rule instead of an error page. The nice thing here is the security team has provided reusable client code that makes this super easy.
An important detail here is that the security team is adding telemetry to how often `savedObjects.resolve()` is called (so they eventually know when we can safely remove it) and we need to _only_ call this API when we need to. This means we will have to call `savedObjectsClient.get()` _first_ and if that returns a 400 not found, then we call the new `.resolve()` API.
Either way, we'll need to add a new HTTP route because we cannot leverage the saved objects client in the browser because `alert` and `action` are hidden types. We will need a new route that either simply calls `savedObjectsClient.resolve()` or the route itself can call `savedObjectsClient.get()`, handle the 400, then return `savedObjectsClient.resolve()`.
See [the PoC PR](https://github.com/elastic/kibana/pull/107611) for more insight into how this might work. | 0easy
|
Title: [alerting] [index patterns] Removal of IIndexPattern type
Body: See https://github.com/elastic/kibana/issues/107220. | 0easy
|
Title: [Alerting] Add warnings for usage of pre-defined ID in rule creation API in non-default spaces.
Body: We currently allow users to pass in a pre-defined ID when creating rules via to the API. Due to breaking changes to make rule SOs share-capable (in which SO IDs will be regenerated), we would like to start logging warnings when this feature is used during rule creation in non-default spaces.
- Add a warning response header when calling the create rule HTTP API with a custom ID in the non-default space
- Add a warning in our docs about using pre-defined ID with the create rule HTTP API in the non-default space (7.12 to 7.16 docs)
- Add a server log warning when calling the create rule HTTP API with a custom ID in the non-default space | 0easy
|
Title: [APM] Convert camelCase to snake_case for all files and folders
Body: Kibana is using snake_case for all files and folders but APM has a mixed filenaming scheme. APM should use snake_case like the rest of Kibana.
APM has disabled Kibana's filename check. This check should be re-enabled again.
**Steps**
1. Remove/uncomment the following lines to re-add case-checking of APM files:
https://github.com/elastic/kibana/blob/9d42d670e1676a622d7ee7d4d564c4d7dc3973e8/src/dev/precommit_hook/casing_check_config.js#L60-L64
2. Run `node scripts/check_file_casing.js` from kibana root
3. Fix all the filename errors listed in the step above
4. Run `node scripts/check_file_casing.js` again to verify all errors are fixed
**Notes**
If you're on a platform with a case-insensitive filesystem (macOS by default), you'll need to have
```
[core]
ignorecase = false
```
in .git/config.
If you're _just_ changing the casing of a filename (from upper to lowercase) you may need to use `git mv -f` to rename the files and make sure github notices.
**Tasks**
- [x] `x-pack/plugins/apm/public/**/*`
- [x] `x-pack/plugins/apm/scripts/**/* ` | 0easy
|
Title: [Integrations UI] Improve Tests for Agent Enrollment Flyout
Body: Ref #103554
Some tests for the [Agent Enrollment flyout ](https://github.com/elastic/kibana/blob/master/x-pack/plugins/fleet/public/components/agent_enrollment_flyout/agent_enrollment_flyout.test.tsx)are currently being skipped because the `step` component passed into the flyout are not first-class React components. We should update this implementation and improve the tests to assert that the expected steps appear in the UI as appropriate. This should include test cases for custom "View Data" step content, as well as UI Extensions that entirely replace the final step of the flyout, such as APM's implementation. | 0easy
|
Title: [APM] Set preference to any value for APM searches
Body: If the APM indices have replica shards, the shard request cache hit ratio goes down, as each shard (primary/replica) has its own cache. If we set `preference` to any value, ES will ensure requests are routed to the same shard if possible, which should result in a better cache hit ratio.
See:
https://www.elastic.co/guide/en/elasticsearch/reference/current/search-shard-routing.html | 0easy
|
Title: [8.0 only] Remove `xpack.task_manager.index` setting
Body: See https://github.com/elastic/kibana/issues/101964. | 0easy
|
Title: Remove `kibana.index`, `xpack.reporting.index`, `xpack.task_manager.index` settings
Body: As discussed in https://github.com/elastic/kibana/issues/82020, we will be removing the following settings in 8.0:
- [x] `kibana.index` - https://github.com/elastic/kibana/pull/112773
- [x] `xpack.reporting.index` - https://github.com/elastic/kibana/pull/113803
- [x] `xpack.task_manager.index` - https://github.com/elastic/kibana/issues/102076
These settings have already been deprecated in 7.x; however, they have not been removed from `master`.
The Observability team currently takes advantage of these settings to allow their developers to use a multi-tenant Cloud environment. They are actively testing a few approaches to do so, but haven't implemented one yet. As a result, we should delay removing these settings until they have a path forward.
Blocked by https://github.com/elastic/observability-test-environments/issues/915 | 0easy
|
Title: [Alerting] Move `kbn-action` tool to the Kibana repository.
Body: To make Alerting framework [support enablement](https://github.com/elastic/kibana/issues/83484) easier and simpler we often use `kbn-action` tool, designed by @pmuellr, which is located under repository https://github.com/pmuellr/kbn-action.
Based on the discussion with @elastic/kibana-alerting-services team and @kobelb we decided to move this tool to be under the Kibana repository. This will help us to maintain it and keep updated according to the framework changes.
In addition, we need to update `kbn-action` to support the new terminology API for Rules and Connectors.
Update Alerting Troubleshooting [docs](https://www.elastic.co/guide/en/kibana/master/alerting-troubleshooting.html) | 0easy
|
Title: [Telemetry] Move telemetry-tools fixtures into package
Body: The telemetry tools package uses fixtures in `src/fixtures/telemetry_collectors`.
These fixtures are using the telemetry plugin `collectorSet` thus why the files lived inside `src` rather than inside `packages/kbn-telemetry-tools`.
Since the tool parses the files statically we can `@ts-ignore` the fixture files and move them into the package.
The `telemetry_collectors` folder is the only folder in `src/fixtures/` so moving it makes sense to reduce surface area of `src`. | 0easy
|
Title: [Actions] Remove deprecated action configuration options
Body: In the [PR](https://github.com/elastic/kibana/pull/100179) we deprecated the TLS related actions configuration options:
1.`xpack.actions.rejectUnauthorized` in favor of the new option `xpack.actions.tls.verificationMode`
2.`xpack.actions.proxyRejectUnauthorizedCertificates` in favor of the new option `xpack.actions.tls.proxyVerificationMode`.
3.`xpack.actions.customHostSettings.tls.rejectUnauthorized` in favor of the new option `xpack.actions.customHostSettings.tls.verificationMode`.
Deprecated options should be removed from the actions configuration, `src/dev/build/tasks/os_packages/docker_generator/resources/base/bin/kibana-docker` and cloud.
This can be done when telemetry from https://github.com/elastic/kibana/issues/108716 shows usage is below 1% or deprecated for > 2.5 years. | 0easy
|
Title: [Alerting] Remove top-level rule fields passed into rule executors
Body: With [this PR](https://github.com/elastic/kibana/pull/99819), we are passing rule information into the rule executors encapsulated in a rule object. Prior to this, we were passing several rule fields as top level fields into the executor, specifically `name`, `tags`, `createdBy` and `updatedBy`. We should audit the rule executors for usage of these top level fields, switch those usages to using the `rule` object, and remove the top level fields from being passed into the executors. | 0easy
|
Title: Mark IIndexPattern as deprecated and add removeBy
Body: As per https://github.com/elastic/kibana/discussions/99462#discussioncomment-730982, IIndexPattern is deprecated (and possibly `IFieldType` too?).
If an `@deprecated` flag is added into the description with an `@removeBy` we can start getting folks off of it and a path for removal. Usage would also show up in the deprecated API list [here](https://elasticdocstest.netlify.app/kibana-dev-docs/deprecated-api-list). | 0easy
|
Title: Cases plugin needs to export some more types publically
Body: Run `node scripts/build_api_docs --stats exports --plugin cases` to see the list:
```
7 referenced API items not exported
┌─────────┬───────────────────────────────────────────┬──────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│ (index) │ Not exported source │ references │
├─────────┼───────────────────────────────────────────┼──────────────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ 0 │ 'cases-public-AllCasesProps' │ 'cases-def-public.CasesUiStart.getAllCases, cases-def-public.CasesUiStart.getAllCases' │
│ 1 │ 'cases-public-AllCasesSelectorModalProps' │ 'cases-def-public.CasesUiStart.getAllCasesSelectorModal, cases-def-public.CasesUiStart.getAllCasesSelectorModal' │
│ 2 │ 'cases-public-CaseViewProps' │ 'cases-def-public.CasesUiStart.getCaseView, cases-def-public.CasesUiStart.getCaseView' │
│ 3 │ 'cases-public-ConfigureCasesProps' │ 'cases-def-public.CasesUiStart.getConfigureCases, cases-def-public.CasesUiStart.getConfigureCases' │
│ 4 │ 'cases-public-CreateCaseProps' │ 'cases-def-public.CasesUiStart.getCreateCase, cases-def-public.CasesUiStart.getCreateCase' │
│ 5 │ 'cases-public-RecentCasesProps' │ 'cases-def-public.CasesUiStart.getRecentCases, cases-def-public.CasesUiStart.getRecentCases' │
│ 6 │ 'cases-server-CasesClient' │ 'cases-def-server.CaseRequestContext.getCasesClient' │
```
Without exporting these types publically they won't show up in our API documentation:
<img width="844" alt="Screen Shot 2021-05-06 at 4 35 40 PM" src="https://user-images.githubusercontent.com/16563603/117362490-53866000-ae89-11eb-802e-1d503de55294.png">
| 0easy
|
Title: Bfetch plugin is has two types that need to be exported bc they are being used in the public API
Body: Run `node scripts/build_api_docs.js --plugin bfetch --stats exports` to see a list of all types that need to be exported.
```
info 2 referenced API items not exported
┌─────────┬────────────────────────────────────────────────┬──────────────────────────────────────────────────────────┐
│ (index) │ Not exported source │ references │
├─────────┼────────────────────────────────────────────────┼──────────────────────────────────────────────────────────┤
│ 0 │ 'bfetch-public-FetchStreamingParams' │ 'bfetch-def-public.BfetchPublicContract.fetchStreaming' │
│ 1 │ 'bfetch-public-StreamingBatchedFunctionParams' │ 'bfetch-def-public.BfetchPublicContract.batchedFunction' │
└─────────┴────────────────────────────────────────────────┴──────────────────────────────────────────────────────────┘
```
Without exporting these types, they won't show up in our API documentation:
<img width="789" alt="Screen Shot 2021-05-06 at 4 11 27 PM" src="https://user-images.githubusercontent.com/16563603/117359701-c2fa5080-ae85-11eb-833d-243bdfcc0848.png">
| 0easy
|
Title: remove any types from core public plugin API
Body: Core team currently has the most `any` types in the public API. Let's get that down!
<img width="1028" alt="Screen Shot 2021-05-06 at 2 14 29 PM" src="https://user-images.githubusercontent.com/16563603/117346871-5461c680-ae76-11eb-92ee-8d98178511ff.png">
Run
`node scripts/build_api_docs --stats any --plugin core`
for a full list with links of every item:
<img width="1378" alt="Screen Shot 2021-05-06 at 2 37 36 PM" src="https://user-images.githubusercontent.com/16563603/117348783-9ee44280-ae78-11eb-9291-09aa485fa882.png">
| 0easy
|
Title: Remove `any` from data plugin's public API
Body: `any` type is always a bad idea, but particularly in a public API item, and even _more_ so if there are no comments.
<img width="658" alt="Screen Shot 2021-05-06 at 2 11 28 PM" src="https://user-images.githubusercontent.com/16563603/117345864-0dbf9c80-ae75-11eb-9b9b-954faa4256af.png">
Data plugin has quite a few
<img width="1028" alt="Screen Shot 2021-05-06 at 2 14 29 PM" src="https://user-images.githubusercontent.com/16563603/117346621-faf99780-ae75-11eb-935a-9be9ba80d2ec.png">
Run `node scripts/build_api_docs --stats any --plugin data` to get a list of every specific API item:
<img width="1646" alt="Screen Shot 2021-05-06 at 2 19 02 PM" src="https://user-images.githubusercontent.com/16563603/117346648-06e55980-ae76-11eb-88a7-f8fdc200e96e.png">
| 0easy
|
Title: Remove `export *` pattern from Data plugin
Body: `export *` makes it easy for a developer to accidentally expose functionality on a plugin's public API, and so should be avoided in favor of explicit exports.
Looks like `data/common/index.ts` has quite a few:
https://github.com/elastic/kibana/blob/master/src/plugins/data/common/index.ts#L9 | 0easy
|
Title: [Alerts][Actions] Find out an automated way to define alerts and actions types for telemetry `byTypeSchema`
Body: Currently `byTypeSchema` is defined manually and not documented to be required to fill for creating new alert and action types.
https://github.com/elastic/kibana/blob/23ce8dcfdb12be92e133242205b779377f5d8d43/x-pack/plugins/alerting/server/usage/alerts_usage_collector.ts#L12-L20
https://github.com/elastic/kibana/blob/f44916b6aabb5ffb519a6953d2bd674415550cc5/x-pack/plugins/actions/server/usage/actions_usage_collector.ts#L12-L20
By that reason, we have now missing telemetry for a new alert types:
- geo-containment
- es-query
and for action types:
- teams
We need to find the way how this schema could be defined automatically for all alert and action types, or at least to provide a documentation where this will be a requirement for defining a new alert/action type.
Related [issue](https://github.com/elastic/infra/issues/27403) | 0easy
|
Title: [ES Client][Alerting] Follow up issue for clean up TS annotations added during migrating alerting plugins to use the new type definitions.
Body: This issue is a follow up request on the changes maded in alerts, actions, stack_alerts, event_log, task_manager, triggers_actions_ui plugins by the [PR](https://github.com/elastic/kibana/pull/83808)
The action is needed is go over the added annotations for possible TS errors and update our plugins code to remove it.
Example:
`// @ts-expect-error not compatible response type`
In addition we should replace all imports like:
```
// eslint-disable-next-line @kbn/eslint/no-restricted-paths
import { elasticsearchClientMock } from '../../../../../../src/core/server/elasticsearch/client/mocks';
```
with the unrestricted path `src/core/server/mocks path` and exposed type `elasticsearchServiceMock` which contains all methods which we use from elasticsearchClientMock | 0easy
|
Title: [Telemetry] Enforce use of `import type` and `export type`
Body: Let's use `import/export type` where it applies. This is recommended for better bundle size, but also for not publicly exposing classes we don't want external services to use (but we still want to share their types).
- [x] packages/kbn-analytics/**
- [x] packages/kbn-telemetry-tools/**
- [x] src/plugins/kibana_usage_collection/**
- [x] src/plugins/usage_collection/**
- [x] src/plugins/telemetry/**
- [x] src/plugins/telemetry_collection_manager/**
- [x] src/plugins/telemetry_management_section/**
- [x] x-pack/plugins/telemetry_collection_xpack/**
We can do this as a follow-up to #92221 | 0easy
|
Title: [Telemetry] Remove uses of `any` in the Telemetry & Usage Collection plugins and libraries
Body: Let's remove (or minimize as much as possible) the use of `any` from all the code related to Telemetry and Usage Collection plugins:
- [x] packages/kbn-analytics/** (#98338)
- [x] src/plugins/kibana_usage_collection/** (#98338)
- [x] src/plugins/usage_collection/** (#98338)
- [x] src/plugins/telemetry/** (#98338)
- [x] src/plugins/telemetry_collection_manager/** (#98338)
- [x] src/plugins/telemetry_management_section/** (#98338)
- [x] x-pack/plugins/telemetry_collection_xpack/** (#98338)
- [ ] packages/kbn-telemetry-tools/**
Once a path is completed, let's add the rule `"@typescript-eslint/no-explicit-any": "error"` in `.eslintrc.js` | 0easy
|
Title: [Security Solution][Enhancement]Common "Save" & "Save & Close" button under Connector Run Tab causing confusion
Body: **Describe the bug**
[Enhancement]Common "Save" & "Save & Close" button under Connector Run Tab causing confusion
**Build Details:**
Platform: Production
Version: 7.12.0 BC1
Commit:038fd51b0474c96e4d3ae58db1308c5f8aba71bd
Build: 38938
**Browser Details**
All
**Preconditions**
1.Elastic Cloud environment having version 7.12.0 should be available.
2.Any External Connector should be present on above environment.
**Steps to Reproduce**
1. Navigate to Kibana and Select Management App from the left navigation panel.
2. Go to Stack Mangement.
3. Go to Alerts and Actions > Connectors Tab
4. Click on existing created external connector and go to 'TEst' Tab
5. Observed the two of the common button actually meant for 'Configuration Tab' are showing under 'Test' tab also even having no requirement of those buttons there.
In Addition clicking on those buttons have no response , which raises a doubt/confusion to a user for the purpose of those or those buttons are not working.
**Actual Result**
Enhancement : Common "Save" & "Save & Close" button under Connector Run Tab causing confusion
**Expected Result**
Common "Save" & "Save & Close" button should be present only under there respective required and working tab i.e "Configuration" Tab and not to shown under "Test" Tab
**Whats Working**
- Save and Save and Close both button are working fine under "Configuration" Tab
**Whats Not Working**
- N/A
**Screenshots**
**logs**
N/A
| 0easy
|
Title: [Lens] Use index pattern service to load list
Body: Part of https://github.com/elastic/kibana/issues/91715
In
https://github.com/elastic/kibana/blob/master/x-pack/plugins/lens/public/indexpattern_datasource/loader.ts#L338-L342
the list of index patterns is loaded via saved objects client.
This should be replaced by using the `getIdsWithTitle` method of the index pattern service. | 0easy
|
Title: [Task Manager] Old `performance` make code harder to maintain
Body: Throughout Task Manager we have api calls made to the `performance` module, consumed by the `plugin_api_perf` test suite.
Now that we have the Health monitoring endpoint and have made great strides in cloud based perf testing - I feel it's time top remove this suite and all the `performance` api calls, as they add a lot of noise in the TM code base that makes it harder to maintain.
Additionally, these hooks are based on how TM worked a year ago, but aren't quite as useful now that the internal logic has changed. If we don't remove them, we should probably rethink how they are laid out to make them more useful. | 0easy
|
Title: [event log] restructure route paths to avoid path parameters at "beginning" of the urls
Body: While reviewing PR https://github.com/elastic/kibana/pull/89681, I happened to notice that are URLs are not structured that great.
```js
const BASE_EVENT_LOG_API_PATH = "/api/event_log"
```
https://github.com/elastic/kibana/blob/b058f7852b74fd22f396c367dc47cf4a526b4daa/x-pack/plugins/event_log/server/routes/find.ts#L25-L27
https://github.com/elastic/kibana/blob/b058f7852b74fd22f396c367dc47cf4a526b4daa/x-pack/plugins/event_log/server/routes/find_by_ids.ts#L29-L31
(new API in PR, shaped the same as others)
https://github.com/elastic/kibana/blob/53d7f763ae684dfd002c1072ebec18ac022180e2/x-pack/plugins/event_log/server/routes/get_events_summary_by_saved_object_ids.ts#L47-L49
Two problems:
- we've now made the path segment after `/api/event_log` always have to be a `type` value - since it could be anything, we can't ever create a fixed path like `/api/event_log/_do_something`, since `_do_something` could be a SO type
- we may want to have APIs that work over multiple types, but now we can't, since you have to always have a single type as the path segment - I guess you could pass one in on the path, and others via query string, but that's awkward
Seems like we need to reshape these to allow for more URL patterns in the event log in the future.
| 0easy
|
Title: [Alerting] AlertParamsExpressionComponent should only be able to update alert type params
Body: The AlertParamsExpressionComponent is currently passed the `setAlertProperty` prop, which allows it to change alert properties other than alert params. Based on discussion in [this issue](https://github.com/elastic/kibana/issues/89880), it has been decided that this behavior should not be supported. AlertParamsExpression components should be scoped to only allow the component to update the alert params. | 0easy
|
Title: Cleanup alerting RBAC exception code
Body: We've introduced some RBAC exemptions in PR https://github.com/elastic/kibana/pull/75563 and https://github.com/elastic/kibana/pull/77598 to avoid delivering breaking changes in a minor. We should deprecate this code and remove it when telemetry from https://github.com/elastic/kibana/issues/108716 shows usage is below 1% or deprecated for > 2.5 years. | 0easy
|
Title: Move kbn-alert-load tool into Kibana alerting
Body: This issue is to move the [kbn-alert-load](https://github.com/pmuellr/kbn-alert-load) tool into Kibana. | 0easy
|
Title: Alerting Task Runner is heavily coupled with its dependnecies
Body: The AlertingTask Runner does [a lot of things](https://github.com/elastic/kibana/blob/master/x-pack/plugins/alerts/server/task_runner/task_runner.ts) and its tests are heavily coupled with its dependencies.
Mocking out dependencies such as the `createExecutionHandler` function breaks many tests and, as you can see in the tests, there's a lot of setup and dependency mocking throughout.
This makes it hard change the runner and hard to reliably test it.
We should try to break it down into smaller units that are easier to test independently. | 0easy
|
Title: Alerting flyouts don't handle changing props particularly well
Body: Some items from the original issue description have been addressed. Remaining tech debt:
> our components are trying to do too many things at once (_AlertsForm_ is both the form for choosing an AlertType and wraps the behaviour for rendering after a type has been selected), and could be addressed by breaking these down into smaller, more specific, components. This should also help in long term maintenance and sustainability of our codebase.
### Original issue description ###
I've encountered a range of UI bugs caused by the state between `AlertAdd` and `AlertForm` going out of sync.
This seems to be caused by race conditions between the `useState` in these two components and is hard to reliably reproduce (I've been able to reproduce it in Uptime quite regularly due to their unique method of selecting an alertTypeId, but it isn't only there), but is definitely present.
I'd recommend abandoning the show/hide approach, which is error prone, and instead going down a render/dont-reender approach which would mean we don't need to maintain state as often as we do now.
While this is, potentially, less performant if the flyout is constantly created/destoryed, this seems unlikely as we don't expect users to regularly create/modify alerts at a high rate. This means that we might actually find this to be more performant as we would avoid instantiating these components in the first place for most users (unlike the current implementation).
A couple of notes:
1. I've already addressed this in that Connectors flyouts in TriggersActionsUI (we now only render these components when needed) in this PR: https://github.com/elastic/kibana/pull/82126. Making the same change in alerting would be a little harder as it's used throughout solutions - we'll have to work with them to address this.
2. Some of these bugs are hidden from sight because our type checking isn't quite true to reality (for example, [this cast](https://github.com/elastic/kibana/blob/f60abf368a53a4ab285c53868d2d1dd7fb011418/x-pack/plugins/triggers_actions_ui/public/application/sections/alert_form/alert_add.tsx#L49) causes one of these state issues to be hidden). This is often because our components are trying to do too many things at once (_AlertsForm_ is both the form for choosing an AlertType and wraps the behaviour for rendering after a type has been selected), and could be addressed by breaking these down into smaller, more specific, components. This should also help in long term maintenance and sustainability of our codebase. 🎉
| 0easy
|
Title: [Trigger Actions UI] Allow async alert type registration from plugins
Body: **Describe the feature:**
For each alert type to work in alerting UI, each plugin has to register an alert type in start contract, that ends up adding few additional bytes into plugin page load bundle sizes, for example for uptime plugin
` plugins.triggersActionsUi.alertTypeRegistry.register(alertInitializer);
`
Plugin ends up importing all alert types defined and those becomes part of page load bundle, it's not a significant chunk, it contains few constants, strings etc, but if we combine for example 20-30 alert types , it might be worth a try to allow plugins to register a promise, and then when alerting ui is loaded, it loads those alert types async.
`TypeRegisrty,register` function will have to be modified to be acceptable for this proposal.
**Describe a specific use case for the feature:**
Improving page load bundle sizes for each plugin, hence improving kibana performance, less code to load and parse.
| 0easy
|
Title: [Actions] Cleanup hidden action types
Body: The case connector introduced at https://github.com/elastic/kibana/pull/80870 can be used, for the moment, only by the detection engine of Security Solution. For that reason, in the same PR, the case connector is being hidden from the `Alerts and Insights` section inside `Stack management`.
When the case connector is being available across Kibana then the logic should be removed. | 0easy
|
Title: [APM] Make size required for ES search requests
Body: We've noticed some cases where `size` was not properly set, e.g. in the wrong part of the request or entirely missing even though we don't need to use `hits`. We can make `size` required in our search request type `/x-pack/plugins/apm/typings/elasticsearch/index.ts` to enforce it always being set. | 0easy
|
Title: [Lens] Pass the uiSettings default language to filters aggregation
Body: **Describe the feature:**
Can be done after merging https://github.com/elastic/kibana/pull/75635
Right now we use 'kuery' as a default language in filters aggregation, but in the future we want this setting to come from uiSettings. The implementation details can be found below (by Joe):
It looks like there are three cases where the default query is required: When adding a new filter using the add button in the UI (which has access to `data`), when initializing an empty filters array (in buildColumn) and when writing the esaggs config. The one in buildColumn could be omitted because it will be caught by the logic in toEsAggsConfig and in the UI anyway.
To solve it cleanly we could extend `toEsAggsConfig` to also pass in `data`. This is a change scoped in the index pattern datasource. To do so,
https://github.com/elastic/kibana/blob/ddf99b64db371f22f6752adc50648fcf2ff413fb/x-pack/plugins/lens/public/indexpattern_datasource/to_expression.ts#L129
has to be replaced by a generator function to pass in the reference and keep it in a function closure:
```ts
export function getToExpression(data: DataStart) {
return function toExpression(state: IndexPatternPrivateState, layerId: string) {
if (state.layers[layerId]) {
return getExpressionForLayer(
state.indexPatterns[state.layers[layerId].indexPatternId],
state.layers[layerId].columns,
state.layers[layerId].columnOrder,
data
);
}
return null;
}
}
```
Then it can be passed through like this:
https://github.com/elastic/kibana/blob/e82e53655d97763bbb8cbf48297d30f5e1b59ffe/x-pack/plugins/lens/public/indexpattern_datasource/indexpattern.tsx#L193
https://github.com/elastic/kibana/blob/ddf99b64db371f22f6752adc50648fcf2ff413fb/x-pack/plugins/lens/public/indexpattern_datasource/to_expression.ts#L129
https://github.com/elastic/kibana/blob/ddf99b64db371f22f6752adc50648fcf2ff413fb/x-pack/plugins/lens/public/indexpattern_datasource/to_expression.ts#L24
https://github.com/elastic/kibana/blob/fdbf1ae7199f06c7511b7aae3e709d2c4dd0f430/x-pack/plugins/lens/public/indexpattern_datasource/operations/definitions/filters/filters.tsx#L144
https://github.com/elastic/kibana/blob/fdbf1ae7199f06c7511b7aae3e709d2c4dd0f430/x-pack/plugins/lens/public/indexpattern_datasource/operations/definitions/filters/filters.tsx#L154
For `buildColumn` it's pretty similar - we call it in the dimension panel and popover editor (which have access to `data` already, so that's easy) and in suggestions. For suggestions, we would need to do the same thing as described above for `toEsAggsConfig`.
| 0easy
|
Title: [Alerting] alerts client find method returns partial alerts, but says it returns full alerts
Body: Currently, the alertsClient `find()` API supports a `fields` option which allows callers to pick which fields they want returned in the alert. So, the result can essentially be a partial alert, not a "full" alert. However the signature on the method indicates it returns full alerts.
It seems like we should do _something_ here, some thoughts on what that would be:
- change the result of the `find()` to be a `Partial<SanitizedAlert[]>` instead of `SanitizedAlert[]`
- remove the `fields` parameter, so you always get a full alert
The second wouldn't make sense if the underlying ES call doesn't actually return fields that aren't specified in the fields parameter, but I'm guessing it does (perhaps the SO client does that filtering as well??). I _think_ the result we're getting back from the SO is essentially the full alert (the `_source` field), so the only reason to filter it out in the API is to reduce memory / network usage for the caller. And I'm guessing there's not much savings there. If that's true, just removing the ability to filter with `fields` seems like it would make sense, and probably makes it simpler for clients, since otherwise they would to deal with a `Partial<Alert>`.
Some relevant snippets:
https://github.com/elastic/kibana/blob/d932830218c00c6548957c0954fcc4ae550352cb/x-pack/plugins/alerts/server/alerts_client.ts#L84-L98
https://github.com/elastic/kibana/blob/d932830218c00c6548957c0954fcc4ae550352cb/x-pack/plugins/alerts/server/alerts_client.ts#L104-L109
https://github.com/elastic/kibana/blob/d932830218c00c6548957c0954fcc4ae550352cb/x-pack/plugins/alerts/common/alert.ts#L49
https://github.com/elastic/kibana/blob/d932830218c00c6548957c0954fcc4ae550352cb/x-pack/plugins/alerts/common/alert.ts#L27-L47
finally, here's the code in `find()` that deletes properties not in `fields` - note the cast to `RawAlert` of the `pick()` result:
https://github.com/elastic/kibana/blob/d932830218c00c6548957c0954fcc4ae550352cb/x-pack/plugins/alerts/server/alerts_client.ts#L356-L364
| 0easy
|
Title: Alerting services deprecated configuration cleanup
Body: Below is a list of deprecated configs that should be cleaned up when telemetry from https://github.com/elastic/kibana/issues/108716 shows usage is below 1% or deprecated for > 2.5 years.
- Remove deprecated `xpack.actions.whitelistedHosts` from #76325
- Remove deprecated `xpack.alerts.healthCheck` from https://github.com/elastic/kibana/pull/92898
- Remove deprecated `xpack.alerts.invalidateApiKeysTask.interval` from https://github.com/elastic/kibana/pull/92898
- Remove deprecated `xpack.alerts.invalidateApiKeysTask.removalDelay` from https://github.com/elastic/kibana/pull/92898 | 0easy
|
Title: [Fleet] Remove timestamp from component template
Body: ES fixed a bug here: https://github.com/elastic/elasticsearch/issues/58956 that required us to include the timestamp field in component templates that defined a mapping section here: https://github.com/elastic/kibana/blob/0b16688a241d3c868ba54b404861f9b71adfe792/x-pack/plugins/fleet/server/services/epm/elasticsearch/template/install.ts#L197
https://github.com/elastic/kibana/blob/0b16688a241d3c868ba54b404861f9b71adfe792/x-pack/test/fleet_api_integration/apis/epm/install_overrides.ts#L67
This can now be removed as the bug is fixed. | 0easy
|
Title: [APM] Endpoints without API tests
Body: ## How to run API tests
The tests are separated in two suites by license: "trial" (similar to Gold+ license) and "basic". To run the tests you must start the server first, and then run the test runner.
**Basic**
```
# Start server
node scripts/functional_tests_server --config x-pack/test/apm_api_integration/basic/config.ts
# Run tests
node scripts/functional_test_runner --config x-pack/test/apm_api_integration/basic/config.ts
```
The API tests tests for "basic" are located in [x-pack/test/apm_api_integration/basic/tests](https://github.com/elastic/kibana/tree/master/x-pack/test/apm_api_integration/basic/tests).
**Trial**
```
# Start server
node scripts/functional_tests_server --config x-pack/test/apm_api_integration/trial/config.ts
# Run tests
node scripts/functional_test_runner --config x-pack/test/apm_api_integration/trial/config.ts
```
The API tests tests for "trial" are located in [x-pack/test/apm_api_integration/trial/tests](https://github.com/elastic/kibana/tree/master/x-pack/test/apm_api_integration/trial/tests).
For debugging access Elasticsearch on http://localhost:9220` (elastic/changeme)
## Endpoints in need for tests
**Services overview page `/app/apm#/services`:**
- [x] `/api/apm/services`
**Transactions overview page `/app/apm#/services/{service.name}/transactions`:**
- [x] `/api/apm/services/{service.name}/agent_name`
- [x] `/api/apm/services/{service.name}/transaction_groups/charts`
- [x] `/api/apm/services/{service.name}/transaction_groups`
- [x] `/api/apm/services/{service.name}/transaction_types`
- [x] `/api/apm/services/{service.name}/annotation/search`
- [x] `/api/apm/services/{service.name}/transaction_groups/breakdown`
- [ ] `/api/apm/services/{service.name}/node/{serviceNodeName}/metadata`
- [x] `/api/apm/services/{serviceName}/transaction_groups/error_rate`
- [x] `/api/apm/services/{serviceName}/transaction_groups/avg_duration_by_browser`
**Transactions details page `/app/apm#/services/{service.name}/transactions/{transaction.name}`:**
tbd
**Traces overview page (`app/apm#/traces`)**
- [x] `/api/apm/traces`
- [ ] `/api/apm/traces/{traceId}`
**Transaction**
- [ ] `/api/apm/transaction/{traceId}`
**Errors overview page (`/app/apm#/services/{service.name}/errors`)**
- [ ] `/api/apm/services/{service.name}/errors/distribution`
- [ ] `/api/apm/services/{service.name}/errors`
**Errors details page (`/app/apm#/services/{service.name}/errors/{error.id}`)**
- [ ] `/api/apm/services/{service.name}/errors/{error.id}`
- [ ] `/api/apm/services/{service.name}/errors/rate?groupId={error.id}`
- [ ] `/api/apm/services/{service.name}/errors/distribution?groupId={error.id}`
**Security**
- [ ] `/api/apm/security/indices_privileges`
**Metrics page (`/app/apm#/services/{service.name}/metrics`)**
- [ ] `/api/apm/services/{service.name}/metrics/charts`
**APM Indicies**
- [ ] `/api/apm/settings/apm-index-settings`
**Custom links**
- [ ] `/api/apm/settings/custom_links/transaction`
**Anomaly detection**
- [x] `/api/apm/settings/anomaly-detection`
- [x] `/api/apm/settings/anomaly-detection/jobs`
- [ ] `/api/apm/settings/anomaly-detection/environments`
**Service Map**
- [x] `/api/apm/service-map`
- [ ] `/api/apm/service-map/service/{serviceName}`
**Service nodes**
- [ ] `/api/apm/services/{serviceName}/serviceNodes`
**UI Filters**
- [ ] `/api/apm/ui_filters/environments`
- [ ] `/api/apm/ui_filters/local_filters/metrics`
- [ ] `/api/apm/ui_filters/local_filters/errorGroups`
- [ ] `/api/apm/ui_filters/local_filters/traces`
- [ ] `/api/apm/ui_filters/local_filters/services`
**index pattern**
- [ ] `/api/apm/index_pattern/static`
- [ ] `/api/apm/index_pattern/dynamic`
- [ ] `/api/apm/index_pattern/title`
**Observability dashboard**
- [x] `/api/apm/observability_dashboard/has_data`
- [x] `/api/apm/observability_dashboard` | 0easy
|
Title: AlertClient.find returns an incorrect Type when the `fields` options is used
Body: the `find` api claims to return `SanitizedAlert` types, but this is not true when the `field` options is used to narrow down the returned fields.
We should address as it could cause mistakes in `getAlertFromRaw` which could in theory break `find` when `fields` is used.
At the moment we're only aware of a use of `fields` in siem, but we don't know who might rely on this in their own implementations in the future.
This can be cleaned up when telemetry from https://github.com/elastic/kibana/issues/108716 shows usage is below 1% or deprecated for > 2.5 years after deprecating legacy terminology usage. | 0easy
|
Title: Core HttpHandler and fetching any -> unknown type
Body: I think all of these default `any` types should be `unknown`s: https://github.com/elastic/kibana/blob/master/src/core/public/http/types.ts#L283 | 0easy
|
Title: [Tech debt] State container examples
Body: Looks like `http` isn't used here: https://github.com/elastic/kibana/blob/master/examples/state_containers_examples/public/with_data_services/components/app.tsx#L59
I'm not sure the route registered in this example is used anywhere?
| 0easy
|
Title: Convert email body UI component to use EUI's markdown editor
Body: Blocked on https://github.com/elastic/eui/pull/2858 | 0easy
|
Title: [Alerting] disable `any` in the Triggers UI plugin
Body: When we worked on https://github.com/elastic/kibana/pull/64161 we found that the Triggers UI plugin represented about 50% of our use of `any` and addressing this as part of that PR would have made it far bigger so we decided to move it to it's own follow up issue.
This issue should address that so that we no longer use `any` where ever avoidable and require an explicit comment (disabling the ESLint rule + an explanation) where unavoidable. | 0easy
|
Title: [Alerting] action validators should be passed allow-list config utils
Body: While reviewing PR https://github.com/elastic/kibana/pull/63450, and thinking about moving the code that PR adds to the actions plugin to a case-specific plugin, it seems like we need to expose our whitelisting validating checks, which are currently private to the actions plugin.
Here's an example usage:
https://github.com/elastic/kibana/blob/e09e6a2b0b2a295d70e4336b803918f933cacd54/x-pack/plugins/actions/server/builtin_action_types/pagerduty.ts#L121-L135
See the code above that to see how `configObject` is curried onto the validator params.
I think what we should probably do is add an optional `validatorServices` parameter object to all the validators - config, secrets, params (params not technically needed at this point, but might as well be consistent, and maybe needed in the future). Validators that need to whitelist will then just get the APIs needed to do the check in that object, without having to curry. Those that don't, won't need to change. | 0easy
|
Title: [Alerting] disabling `any` within triggers_actions_ui
Body: Follow up from https://github.com/elastic/kibana/issues/64145
complete removal of explicit `any` in Alerting Services code | 0easy
|
Title: [Alerting] TypeScript checking for actionVariable definitions
Body: Let's say we have an alert type defined with:
```js
actionVariables: {
context: [
{ name: 'foo', description: 'Foo' },
{ name: 'bar', description: 'Bar' },
]
}
```
and then we write code to schedule actions like:
```js
alertInstance.scheduleActions('default', {
foo: 1,
baz: 2
})
```
Can we make TypeScript throw an error about the missing `bar` and the present `baz`? Either with full type inferencing or just disallowing us from using `any` types in these functions. | 0easy
|
Title: Should we use minLength: 1 on string validations in alerting?
Body: It was brought up in this PR https://github.com/elastic/kibana/pull/60468#pullrequestreview-376927186 that there may be other places within the alerting framework that could benefit from requiring non-empty strings. | 0easy
|
Title: [Alerting] Concentrate public Alerting Framework components in the alerting plugin
Body: There is some overlap between *alerting* and *triggers_actions_ui* and we should concentrate the client side APIs of the framework itself under the alerting plugin.
For example, there are two alert_api.ts across both plugins, and they can be merged (they aren't exactly the same, but could reside together). | 0easy
|
Title: [Alerting] move always-available action variables to alerting/common
Body: In PR https://github.com/elastic/kibana/pull/59756 action variables were added as a new property of alert types. There are a handful of "always available" action variables such as `alertId` and `alertName`. The definitions of these are duplicated in the `alerting` and `triggers_actions_ui` plugins:
https://github.com/elastic/kibana/blob/ce1836b2af68a49b347f10e8cbdd92f9998ad616/x-pack/plugins/alerting/server/task_runner/transform_action_params.ts#L38-L47
https://github.com/elastic/kibana/blob/ce1836b2af68a49b347f10e8cbdd92f9998ad616/x-pack/plugins/triggers_actions_ui/public/application/lib/action_variables.ts#L27-L42
To prevent these from diverging in the future, we should move them to the alerting/common folder, and then reference them in the files above from there. | 0easy
|
Title: [Discuss][Alerting] Reorganise plugin structure
Body: Seems we should do some clean up.
# Discussion points:
## Index Threshold & ES Query builtin ruleType
The Index Threshold & ES Query RuleType's UI resides in Triggers UI instead of the stackAlerts plugin. My thinking is that builtin plugins should have both a server and a public dir, as that means the UI and BE of each builtin will reside side by side.
## Which code belongs in the Alerting plugin and which in TriggersActionsUI plugin
Figure out when to use an independent plugin and when to extend the alerting plugin. Currently we have two plugins `alerting` and `triggers_actions_ui`.
At the moment `alerting` only has a server side plugin for the api used by consumers of the Alerting framework, while `triggers_actions_ui` has both server and public plugins, as is includes the alerts UI and the servers side code used to power that UI.
With that in mind, the ViewInApp PR (https://github.com/elastic/kibana/pull/58997) introduced a `public` plugin to the `alerting` as it extends the API powering the Alerting Framework with client side navigation logic, which feels independent of the `triggers_actions_ui` (even though it is _used_ by `triggers_actions_ui`). We need to decide whether it belongs in the `triggers_actions_ui` (meaning plugins who want to use it will have to depend on both `triggers_actions_ui` and `alerting`) or whether it belongs under `alerting` (meaning plugins that want to use Alerts within their plugin only need to familiarise and rely on `alerting`). | 0easy
|
Title: [alerting] event log errors when multiple event log resources
Body: I noticed a wall of text when starting Kibana, and happening to have an old `.kibana-event-log` in my indices - we've since moved to using versions in the index name (eg, `.kibana-event-log-8.0.0'): https://gist.github.com/pmuellr/45a18addc7eb71a5634e673ee33a4197
~~Additionally, but may not be related, was unable to load the Kibana web ui, probably need to `yarn kbn bootstrap`, this from running some code I'd just pulled from GH - nothing logged in Kibana.~~ Narrator: it was `yarn kbn bootstrap`.
This isn't really a bug in our code, per-se, since we're not expecting people to have multiple event log indices in a cluster, that aren't all managed by the event log plugin. But perhaps it's an example of what could go wrong in a customer environment.
Basically, just logging this to ponder over it, wondering how bad is this (I didn't investigate), and if it's not great, what can we do to remediate. | 0easy
|
Title: Should we rename lastScheduledActions?
Body: Follow up from PR comment: https://github.com/elastic/kibana/pull/56625#discussion_r375273107 | 0easy
|
Title: [APM] Change get transaction by trace id url
Body: The current API to get the transaction based on the trace ID is `/api/apm/transaction/TRACE_ID`, but based on restful principals this is incorrect.
To make it consistent we should change it to `/api/apm/traces/{traceId}/root_transaction`, and move it to https://github.com/elastic/kibana/blob/master/x-pack/legacy/plugins/apm/server/routes/traces.ts#L19 | 0easy
|
Title: [APM] Use `ProcessorEvent` constants everywhere instead of literals
Body: We have a `ProcessorEvent` enum:
https://github.com/elastic/kibana/blob/d1de029b8873e29c35df4bf6ce95f987efd83874/x-pack/legacy/plugins/apm/common/processor_event.ts#L7-L12
This should be used instead of string literals:
```js
// bad
term: { [PROCESSOR_EVENT]: 'transaction' }
// good
term: { [PROCESSOR_EVENT]: ProcessorEvent.transaction }
```
**Examples where `ProcessorEvent` is used:**
https://github.com/elastic/kibana/blob/d1de029b8873e29c35df4bf6ce95f987efd83874/x-pack/legacy/plugins/apm/server/lib/transactions/avg_duration_by_browser/fetcher.ts#L30
**Examples where `ProcessorEvent` should be used:**
https://github.com/elastic/kibana/blob/d1de029b8873e29c35df4bf6ce95f987efd83874/x-pack/legacy/plugins/apm/common/projections/services.ts#L34
https://github.com/elastic/kibana/blob/d1de029b8873e29c35df4bf6ce95f987efd83874/x-pack/legacy/plugins/apm/server/lib/index_pattern/get_dynamic_index_pattern.ts#L81-L83
https://github.com/elastic/kibana/blob/d1de029b8873e29c35df4bf6ce95f987efd83874/x-pack/legacy/plugins/apm/server/routes/index_pattern.ts#L28-L30
... and many other places. | 0easy
|
Title: Allow actions to provide ECS data for the event log
Body: With the coming event log, it probably makes sense for actions to provide ECS-compatible data on what they're doing, if they have anything. For instance, any action using an http service could provide data from the [http fieldset](https://www.elastic.co/guide/en/ecs/current/ecs-http.html).
Not sure the best way to provide this ATM. We could either pass an event logger into the action, and then let the action do the logging, or allow the action to return data to be added to the event log document generated for an action execution. | 0easy
|
Title: Remove Boom errors for non API related files
Body: This would mean to catch errors at the API level and convert to proper status code. This is required for new platform and NP best practices. | 0easy
|
Title: Add security tests for es_index built-in action type
Body: This is the only current built-in that should be sensitive to the user associated with executing an action, in that it performs es calls as part of action execution. | 0easy
|
Title: Add createdBy and updatedBy fields to action saved objects
Body: Implementation to be like alerts: https://github.com/elastic/kibana/pull/41389#discussion_r315679918. | 0easy
|
Title: [Monitoring] License expiry_date_in_millis UI test
Body: Follow up from elastic/kibana#21354
We should add a UI test that ensures `expiry_date_in_millis` is used in the UI | 0easy
|
Title: [Meta]: Port away from ILogger
Body: ## Current pull requests
- [x] https://github.com/nextcloud/server/pull/31609
- [ ] https://github.com/nextcloud/server/pull/31943
- [ ] https://github.com/nextcloud/server/pull/31948
- [ ] ...
## Current remaining modules to port in server
- [ ] lib/{private,public}/Files
- [ ] federation
- [ ] encryption
- [ ] federatedfilesharing
- [ ] files_sharing
- [ ] files_trashbin
- [ ] files_external
- [ ] ....
## Current remaining shipped app to port
- [ ] ....
## Current issues
- The background job interface exposes `ILogger` in its public API. We need to deprecate this
- In some places, like the updater. It's helpful to be able to write `ILogger:DEBUG` instead of `2`. We need to look if upstream psr would be interested to add enums for the log levels. | 0easy
|
Title: Fix `vue/no-mutating-props` issues
Body: We badly started implementing some props mutation.
There are usually two cases:
1. The props are injected via root component `propsData`
2. The props are used inside a vue app
## Solution 1
Migrate the propsData (usually initial-states) to the root `.vue` file away from the `.js` entry point
## Solution 2
Use a sync modifier and change the data flow according to vue documentations. Or rethink the way you're using the data and consider switching to events and move the handling methods one component up.
| 0easy
|
Title: Phase out untyped service injection in DI container
Body: For historical reasons we still have aliases for parameters without type hints, like `Logger` for the ``\OCP\ILogger``. As we now type hint everything anyway, I think these aliases can be phased out slowly.
https://github.com/nextcloud/server/blob/2be1a6a04e024c93cda059fafa2b1fa405c86f44/lib/private/Server.php#L680
@blizzz @icewind1991 @nickvergessen @rullzer does that make sense? What could be a possible deprecation path? Document deprecation and remove later? | 0easy
|
Title: Allow to get CookieJar in IClientService
Body: With IClientService you can get a wrapped GuzzleClient.
However you have no way to keep the cookies with our API.
The following would work:
```php
$cookieJar = new \GuzzleHttp\Cookie\CookieJar();
$client = $this->clientService->newClient();
$response = $client->post(
'.../login', [
'body' => [
'name' => $name,
'password' => $password,
],
'cookies' => $cookieJar,
]
);
$response = $client->get(
'.../authed-api', [
'cookies' => $cookieJar,
]
);
```
I think we should add a `newCookieJar()` method to the clientService, to hide the Guzzle CookieJar. I just don't know what to do with the interface, Should we add a dummy around it like with the client? and forward all calls to the internal?
cc @MorrisJobke @rullzer agreed and okay for 16? | 0easy
|
Title: Move core apps backgroudjobs to OCP
Body: Since N15 we finally have an OCP mechanism to handle background jobs.
We should use it properly. | 0easy
|
Title: Settings javascript should move away from our obsolete jquery
Body: There is still a lot of javascript left in https://github.com/nextcloud/server/tree/master/settings/js
+It would be good to either move it to vue (in small steps) or to plain javascript.
## Progress
- [x] Apps
- [x] Users management
- [ ] Admin
- [ ] Basic params
- [ ] Security
- [x] 2FA
- [ ] Sharing
- [ ] Personal
- [ ] Personal info (tracked in #27869)
- [ ] Security
- [ ] Password change
- [x] AuthTokens
- [x] WebAuthn
- [ ] Overview (setup checks) | 0easy
|
Title: Move contactsmenu to vue
Body: The contactsmenu (behind the contact icon on every page on top) should be moved to Vue.
The vue programs from settings are probably a good place to start. | 0easy
|
Title: Fix typo in IUserManager::createUser doc
Body: | 0easy
|
Title: Submit button of login page should not use `id="submit"`
Body: The 57th line of the file `/core/templates/login.php` have a potential bug
```
<input type="submit" id="submit" class="login primary" title="" value="<?php p($l->t('Log in')); ?>" disabled="disabled" />
```
When I run the code `document.querySelector('form[name="login"]').submit()` under login page and an error will be thrown
```
VM78:1 Uncaught TypeError: document.querySelector(...).submit is not a function
at <anonymous>:1:46
```
I can't submit this form because `document.querySelector('form[name="login"]').submit` is not a function but a HTMLElement of the submit button.
Please fix it and thanks!
Version Of Server: v13.0.4 and v14 | 0easy
|
Title: PHP built-in server crashes when performing certain actions on Nextcloud
Body: This is more a curiosity than something that needs to be fixed (as nobody would use the PHP built-in web server to run Nextcloud in the real world, and the bug is triggered by Nextcloud, but it does not look like a Nextcloud bug), but I thought it was worth documenting it.
In the last days the acceptance tests for the Files app have been failing. Specifically, the [_add tags using the dropdown in the details view_](https://github.com/nextcloud/server/blob/15c12eb0270a35fa875e519b04d8378afc6e795d/tests/acceptance/features/app-files.feature#L178) and [_remove tags using the dropdown in the details view_](https://github.com/nextcloud/server/blob/15c12eb0270a35fa875e519b04d8378afc6e795d/tests/acceptance/features/app-files.feature#L206) scenarios. However, if you performed the actions from those scenarios manually on a real web server everything worked fine. Why did they fail on Drone then?
Well, they failed because the PHP built-in web server crashed (SIGSEGV). Really :-P That crash was not related to Drone nor the acceptance tests, though; it can be reproduced with the steps below:
- Create a Docker container for the PHP built-in server
- On a Nextcloud Git directory, checkout the first offending commit with `git checkout 09940bcde6 && git submodule update --recursive`
- Start a Docker container for the Nextcloud server in that directory with `docker run --rm --volume /PATH/TO/THE/NEXTCLOUD/GIT/DIRECTORY/:/var/www/html --volume /var/www/html/data --volume /var/www/html/config --interactive --tty --name php-builtin-server-crash nextcloudci/php7.1:php7.1-16 bash`
- In a different terminal (as the previous one will be in the Bash session inside the container), get the IP of the new container with `docker inspect --format='{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}' php-builtin-server-crash`
- Start the PHP built-in server (all the following commands are performed in the Bash session inside the container)
- Change to the Nextcloud directory with `cd /var/www/html`
- Install Nextcloud with a SQLite database with `php occ maintenance:install --admin-pass=admin`
- Whitelist the IP of the container with `php occ config:system:set trusted_domains 0 --value="IP_OF_THE_CONTAINER"`
- Start the PHP built-in server with `php -S IP_OF_THE_CONTAINER:80 -t .`
- Make the PHP built-in server crash
- Visit _http://IP_OF_THE_CONTAINER_ in a web browser
- Login as _admin_ with password _admin_
- Open the settings, and then the _Workflow_ section
- Create a tag named _tag1_
- Create a tag named _tag2_
- Show the list of tags
When the list of tags is shown the image _http://IP_OF_THE_CONTAINER/core/img/actions/search.svg_ is loaded, and that is what causes the crash. It is not a problem of the image, though; loading any other resource (anything that does not start with _index.php_, like images, CSS files, or even just _http://IP_OF_THE_CONTAINER/core_ or _http://IP_OF_THE_CONTAINER/apps_) causes the crash, but it does not happen if _http://IP_OF_THE_CONTAINER/index.php_ is loaded instead.
Moreover, the crash requires a resource to be loaded immediately after creating two or more tags in a row; it does not happen if just one tag is added and a resource is loaded, or if one tag is added, a resource is loaded, another tag is added, and then another resource is loaded.
Finally, the crash happens when using the Docker images for PHP 7.1 (_nextcloudci/php7.1:php7.1-16_) and PHP 7.2 (_nextcloudci/php7.2:php7.2-11_), but not in PHP 7.0 (_nextcloudci/php7.0:php7.0-19_).
The first offending commit was 09940bcde697df90423cb0b4ecb4e62ba31bfaf7, and it was magically fixed in d40afac1b9e9fe4bb7b64d2c5014a9ba6edb8957.
Most plausible explanation to all this seems to be _A wizard did it_.
| 0easy
|
Title: Headings semantics
Body: @nextcloud/designers I've noticed some issues with semantics that might be worth fixing - notably, headings. This might have to be separated into several issues/PRs:
- [ ] `<h1>` is used twice (it [shouldn't be](http://w3c.github.io/html/sections.html#rank)): once in the NC logo `<h1 class="hidden-visually">Nextcloud</h1>` and once in the `header-appname-container menutoggle` (app name). Possible solutions are changing the tag of the first one or removing the whole link from the Nextcloud logo (it's a weird link anyway, since it targets default app (Files), not Nextcloud itself)
- [ ] because of the :arrow_up: , `<h1>` should never be used in apps - e.g. @nextcloud/news and @nextcloud/deck are breaking this "rule"
- [ ] we have a default (core) style rules for `<h2>` and `<h3>`, but `<h4>` might be useful as well
- [ ] `<h3>` should be used in right sidebar (I guess?) - in which case, we should come up with a good default rule for `#app-sidebar h3`; on a related note, should it actually be an input field, like in @nextcloud/calendar ?
- [ ] default style for subtitle might be useful. @nextcloud/news has `.subtitle` which might be a good starting point (missing a more distinction for links IMO), @nextcloud/mail might apply something similar instead of the `.transparency`
- [ ] the above mentioned apps could use `<header>` [tag](http://html5doctor.com/the-header-element/) to group title and subtitle
Also possibly related to @nextcloud/theming
P.S. Sorry for bringing out such low-level issues :disappointed: | 0easy
|
Title: Proof of concept #5593 - inline SVG icon for icon-contacts
Body: I added `core/img/places/contacts.svg` icon inline, replacing CSS implementation as `background-image`
This would solve some weird css class names and allow for a great flexibility with theming icons.
it should work across all supported browsers.
I also noticed that an almost identical icon is used in dropdown settings menu in lib/private/NavigationManager.php#211 - this solution would work there as well, but it would require changing the toggle method. | 0easy
|
Title: Better handling of inverted icons
Body: My thoughts related to #5450 - which is just one example of problems with icons in NC.
I don't think using `icon-more-white` to display dark variant of the icon is a very elegant approach. I'm wondering if we can come up with a better solution for such situations ( #5450 is only an example). Here are several proposals that we might want to consider in future releases:
1. SVG: use inline SVG as icon source (many pros and cons, I'm not going to list them here), set `fill` in CSS as needed
2. JS: toggle between `icon-more` and `icon-more-white` classes using JS
3. CSS: use CSS `filter: invert(100%)` property, which is quite nicely [supported](http://caniuse.com/#feat=css-filters) already
My personal favourite of those is `fill`, but `filter()` would also be neat. Both of these would allow us to deprecate a considerable amount of existing icons. They would also give us possibility to theme icons (I guess folder icon is the only obvious candidate here).
SVG-based solution would additionally allow us to have icons-within-icons without maintaining a separate asset, such as core/img/filetypes/folder-public.svg - but it might not be feasible just yet :cry:
I would like to hear your thoughts @nextcloud/designers @nextcloud/theming | 0easy
|
Title: JVM Metrics tutorial does not work
Body: **Page**
[Metrics System @ JVM Metrics](https://docs.alluxio.io/os/user/edge/en/operation/Metrics-System.html#jvm-metrics).
**Summary**
The actions specified by the docs are difficult to understand, do not provide a clear step by step instruction plan, and ultimately do not work.
| 0easy
|
Title: ProcessUtils.fatalError() to export critical information before crashing the process
Body: **Is your feature request related to a problem? Please describe.**
It would be good to export some critical information (stack traces, heap dump, etc...) before crashing process due to a fatal error.
**Additional context**
It's hard to reproduce fatal errors and most of the valuable information is gone with the terminated process.
| 0easy
|
Title: Improve the error message when master hostname is not set
Body: **Summary**
When the following message is given, it looks like ZK or HA is required, which is not true
```
Cannot run Alluxio fs shell; Unable to determine master address. Please modify alluxio-site.properties to either set alluxio.master.hostname, configure zookeeper with alluxio.zookeeper.enabled=true and alluxio.zookeeper.address=[comma-separated zookeeper master addresses], or utilize internal HA by setting alluxio.master.embedded.journal.addresses=[comma-separated alluxio master addresses]
```
**Urgency**
Low | 0easy
|
Title: [FEATURE] Filters for hooks
Body: Continuation of #1673
Extend the hook system to allow users to specify the scope of hook functions through a decorator instead of implementing conditions inside the hook function body. This would simplify the filtering process and make it easier to use.
```python
# Filter via custom function
def select_operations(context):
# Or any complex logic here
return context.operation.verbose_name == "PATCH /users/{user_id}/"
# Apply hook only if `select_operations` returned `True`
@schemathesis.hook.apply_to(select_operations)
def filter_query(context, query):
return query["key"] != "42"
# Apply hook to everything except `GET /users/{user_id}/`
@schemathesis.hook.skip_for(method="GET", path="/users/{user_id}/")
def map_path_parameters(context, path_parameters):
path_parameters["user_id"] = 42
return path_parameters
# Simple filter hook for all operations
@schemathesis.hook
def filter_query(context, query):
return query["key"] != "42"
``` | 1medium
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.