This adds an additional check and corresponding error message to the validation of the model memory limit. If the parsing of the memory model limit results in 0 bytes the message will be triggered. This behaviour catches both correctly formatted but invalid strings (like 0mb) and invalid strings (like asdf).
Updates the behaviour of the job validation's model memory limit checks: The success messages success_mml gets only returned if the checks were actually run. This omits the success message for example in the single metric and population wizard.
Inconcistencies between AppState and derived custom states for UI components like mlSelectLimit and mlSelectInterval could trigger errors because the parent of a nested object attribute could become undefined. This PR adds additional checks to state_factory.js's get/set/reset methods to reinitialize itself with AppState and fall back to its default state should it not be present within AppState anymore. The original issues were hard to reproduce consistently using the UI only. However the error could be triggered when creating a malformed URL, like replacing the mlSelectInterval:(interval:(display:Auto,val:auto)) part of the URL to something like mlSelectInterval:(undefined). This PR includes additional unit tests which simulate this behaviour and fail without the additional checks provided by this PR.
* [console_extensions] disable when console is disabled
* [console extensions] has console and enabled
* [console_extensions] fix console_extensions.enabled
* Expand coverage of dashboard tests and decrease time
* Fix timing error when sub urls fail to save from too fast app link clicking
* discover doesn't have breadcrumbs
* Check top nav text so it works on both listing and saved object edit/view pages
* need to do the add panel operations one at a time
* Need both types of input in filter
* Give test data a title
* Remove incorrect and unnecessary comment
* Move data around and get rid of 6_3 specific naming as we will end up migrating the data as we progress
* Remove code accidentally checked in
* [Stats] Add metrics collector and stats API
* uptime_ms in the process namespace
* make uptime_in_millis always equal process.uptime_ms
* fix api integration test
* fix api integration test better
* fix false positive with last change
* change object detection, add fallbacks to return undefined
* Adding basic syntax highlighting for grok expressions
* Use EUI color palette
* Handle regex tokens, escaped and unescaped
* Return token for escaped content
* Add functional test
* Using higher-contrast colors
* Removing comment I used for developing the highlight rules
* Using object destructuring
* Removing unnecessary method
* Extracting constant for reuse
* Keep track of recently-deleted pipelines until Monitoring stops reporting them
* Adding explanatory comment for constant
This reduces the output from the clusters API in the data processing stage.
Mainly, it throws away a lot of unused fields in `elasticsearch.cluster_stats` and `elasticsearch.cluster_state`
cc @rasroh
The documentation states, that after using `yarn start` one should navigate to https://localhost:5601. For me it did not work, since kibana only seems to listen to https if one provides the "--ssl" flag as option (just as described in #### Setting Up SSL).
Thus, I changed the documentation to reflect this fact.
Implement support for the richer style Vega tooltips, per https://github.com/elastic/kibana/issues/17215 request. Uses [Vega tooltip plugin](https://github.com/vega/vega-tooltip) for formatting. Always enabled unless user sets `tooltips=false` flag in the `config.kibana` section of the spec.
![image](https://user-images.githubusercontent.com/1641515/39344487-7abc9eb0-49eb-11e8-89dd-bf562563ae1c.png)
## Test code
```js
{
$schema: https://vega.github.io/schema/vega/v3.json
config: {
kibana: {
tooltips: {
// always center on the mark, not mouse x,y
centerOnMark: true
position: top
padding: 20
}
}
}
data: [
{
name: table
values: [
{
title: This is a long title
fieldA: value of fld1
fld2: 42
}
]
}
]
marks: [
{
from: {data: "table"}
type: rect
encode: {
enter: {
fill: {value: "#060"}
x: {signal: "40"}
y: {signal: "40"}
width: {signal: "40"}
height: {signal: "40"}
tooltip: {signal: "datum || null"}
}
}
}
]
}
```
* update to eui 0.0.44
* keep eui dependency in x-pack
* wrap EuiBasic table column in span to avoid className prop from getting overridden
* remove brittle functional tests that are no longer needed with EUI components
* filter field list instead of chaning table size
* update x-pack snapshots
* convert dashboard listing page to react and EUI
* add jest test for DashboardListing component
* add data-test-subj attributes
* clean up jest test
* hideWriteControls and call to action when no dashboards exist
* pass initial filter to dashboard listing, get functional tests to work
* fix dashboard queries functional tests
* upgraded to EUI 0.0.29 to get defaultFocusedButton fix
* move dashboardListing directive to index
* spacing in if statement
* switch to EuiBasicTable
* pagination
* add sorting
* fix jest test
* handle out of order fetchs
* remove info.gif
* re-instate search functional test
* replace EuiSearchBar with EuiFieldSearch
* fix functional tests
* update snapshot - when code rebased - new EUI version add another prop
* add Edit link to actions column
* [Monitoring/React] Render ES Nodes Listing with Base Controller
Refactors the ES Nodes Listing to use Base Controller and a React component instead of an Angular directive
* remove another obsolete file
Turns out there is a significant problem with how Vega library manages its dependencies. Instead of using a fairly common "^1.2.3" form, it used "1" (major only) almost everywhere. While this works fine if node_modules and yarn.lock are deleted and rebuilt, bumping Vega lib's primary version in our package.json and using yarn install would not touch other subcomponents. This resulted in several 6.2 and 6.3 shipping with the outdated vega dependencies while we thought that we included the latest vega versions The problem has been reported vega/vega#1259 and will hopefully be solved with the next version.
In the mean time, this patch updates all vega-related dependencies in the yarn.lock file.
* [Status] Organize tests into __test__ directories
* custom mock for fs in cgroups test
* work around unhandled rejection errors
* simplify fserror construct
* put fs readFile mockImpl in beforeAll for pretty
* undo hacking the code to suppress unhandled promise rejections
* test files back to sibling of the module
This changes the Advanced Node page for Elasticsearch to recognize the
"write" threadpool, while also incorporating the deprecated "index"
threadpool and renamed "bulk" usage (renamed to "write").
This sums the three writing threadpools together and displays it as a
unified view.
* Update tutorial-define-index.asciidoc
Added a note on how to check for current indices in Elasticsearch, to make it clearer for people that haven't used the previous chapters.
* Update tutorial-define-index.asciidoc