-
Notifications
You must be signed in to change notification settings - Fork 8.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[New Platform] 8.0 Refactor kibana.yml #46705
Comments
Pinging @elastic/kibana-platform |
This one might get tricky, as there are other plugins which use this field to derive their own data which isn't strictly related to saved objects. For example, the Security plugin uses the I think the alerting team is also planning on using this value to derive the name of one of their indices, in order to support multi-tenant setups, and perhaps other features. I'm not sure what domain this would fall under... |
You could definitely argue that |
As part of this refactor it might be worth switching from camel case to snake case convention for consistency across the stack #7444 (thanks @azasypkin) |
Throwing this out there: |
These should probably be moved into the |
For clarity: we are not planning on working on this until after the Platform migration work has been completed (at the earliest). In the meantime, the workaround outlined in #46240 will be the near-term solution. @lukasolson We'll make sure to include the config keys you listed above and it should be very easy to add additional options in the workaround mechanism in the future if we find more. |
I also see a Should the
Seems unused. should probably remove yes
This seems used by multiple plugins. So I guess this should stay there?
Should move to
These are used in should is really moves to data plugin?
Unclear where this should move to, as it's quite an isolated property.
|
Some of these are stack globals, so that should be taken into consideration too. path.data For settings that mean the same thing I'd push for consistency. |
@lukasolson shouldn't it be a part of |
@restrry Yeah, it would make more sense there. |
@pgayvallet I started a separate issue for Kibana app related configs: #54497 @nickofthyme Is working on some of those in #55937
👍
I think it makes sense. Mid-term we can probably refactor input control vis to leave it up to the data fetching layer to set terminate after and timeout. Short-term we could expose it in the contract of the data plugin so other plugins can also make use of it but it's clear the |
@lukeelmers Is the app arch team content with moving |
@flash1293 Thanks for the ping; yes moving those settings to the @rudolf Is there a plan yet for when the other items might be moved, so that we can coordinate? I've created #66085 to track this on the app arch side, but I'm thinking it might make sense to plan these for the same release, since it is a breaking change. |
@lukeelmers We'll have to do this in 8.0 (I'll update the issue to make this more clear) and we will coordinate this with the wider Kibana team to ensure that we include all teams' changes. |
To be clear, we can start migrating this now and use the config deprecations feature to map the old name to the new name. We just can't remove that deprecation translation until 8.0. Ideally, we do these changes earlier than later so sysadmins have more heads up. I don't think it's critical that these deprecations are done at the same time. |
We should move kibana/src/core/server/csp/config.ts Line 28 in 614bde9
|
Should this be moved to an earlier release then? It's currently on our |
8.0 and 7.last are released at the same time, so they're sorta the same roadmap. We need to make sure the deprecation is in the last minor release though. I will move. |
I was more saying that, as we already know that these are going to be moved/deprecated, maybe we should add the deprecations sooner 😄 ? I mean, as a user, having the info that these properties are going to be deprecated is better sooner than later? |
As doing a Kibana-wide renaming of config keys is nowhere on our near-term horizon, we'll go ahead and close this issue for now. We can reopen later if it becomes necessary. |
@lukasolson The data plugin still uses |
Thanks for resurfacing this @TinaHeiligers... Yes, ideally this setting would be in the |
We'd need to follow the new deprecation guidelines before we could remove the old config, meaning the setting would probably linger for a couple years. However, I believe we could still move it to the data plugin starting in 8.1 (too late for 8.0), and just register a deprecation rename for it in the meantime so that users aren't affected. @kobelb Do we need to discuss deprecations before they are added in the first place, or just need sign-off on the actual breaking changes before they happen? |
Deprecations don't need approval. We just need the approval to make the breaking change itself. |
Created a new issue for tracking this last change here: #118250 |
2019-11-19 Updated based on comments
Core's config service uses "paths" to split up
kibana.yml
into sub-config's with their own schema. We want sub-configs to be grouped by domain, but some of the existing config prefix paths don't correspond to any domains in Core.For instance (not a complete list):
migrations.scrollDuration
- The scroll value used to read batches of documents from the source index. Defaults to15m
.migrations.batchSize
- The number of documents to read / transform / write at a time during index migrationsmigrations.pollInterval
- How often, in milliseconds, secondary Kibana instances will poll to see if the primary Kibana instance has finished migrating thekibana.enabled
removedeprecatedkibana.defaultAppId
moved tokibana.disableWelcomeScreen
home
pluginkibana.autocompleteTerminateAfter
move to data pluginkibana.autocompleteTimeout
move to data pluginpath.data
move to be under "kibana"server.uuid
should move outside of theserver
namespaceThis refactoring would be a breaking change so it will have to be introduced in 8.0.0. We should audit all of our configuration options and coordinate this with all teams to ensure that we're satisfied with the end result for all of them.
The text was updated successfully, but these errors were encountered: