Max Fan (481dbda2) at 18 Mar 00:18

... and 1 more commit

Max Fan (6ca9a7e8) at 18 Mar 00:18

What does this MR do and why?

Remove unneeded build arg to with_read_consistency

Ci::BuildTraceChunk#with_read_consistency took a build argument, but that argument was unused in the method body. Calling this method was the only reason Ci::BuildTraceChunkFlushWorker was loading a Ci::Build, so removing the unused argument makes the worker cheaper.

This should help reduce replica lock contention happening in gitlab-com/gl-infra/production#21562

References

gitlab-com/gl-infra/production#21562

MR acceptance checklist

Evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.

👍 I didn't have time to review this MR, but testing it looks good. It added a comment to the broken MR with a link to the new MR

The fix was merged - triggering a new pipeline !227689 (merged)

Thanks for the distinction @nhervas !

I think @cheryl.li and @rutshah want the data to be in tableau for leadership and also to view it themselves? I guess we can first start with Snowflake as well if tableau is tricky to import :)

@nhervas not exactly. All the internal events above each have their own data. For example they have action/property/label metadata: gitlab-org/modelops/applied-ml/code-suggestions/ai-assist!4755 (merged)

Can we import the recent 7 days into tableau so we can filter and view metrics of these internal events and metadata?

sounds good, rubber stamping and merging 🚢

Max Fan (ebcc65a4) at 17 Mar 21:22

... and 1 more commit

Max Fan (0f2df2f9) at 17 Mar 21:22

What does this MR do and why?

Create new Empty State for the Pipeline Editor

This empty state reflects uses the new Visual Language that will be included as part of the update_visual_language feature flag

References

• Issue: #590980
• Rollout Issue for the Feature Flag: #590044

Screenshots or screen recordings

Before	After

How to set up and validate locally

Without the AI Card

1. Enable the update_visual_language feature flag

$ echo "Feature.enable(:update_visual_language)" | gdk rails c

2. Go to a project that has no CI configuration
3. Go to Build -> Pipeline Editor

With the AI Card

1. Enable the update_visual_language feature flag

$ echo "Feature.enable(:update_visual_language)" | gdk rails c

2. Follow the steps to enable Duo in your local GDK
3. Go to a group and ensure Duo is enabled for that group, e.g. with in the FlightJS group

Settings -> General -> GitLab Duo Settings:

If you open Duo and you get a message that there are no credits remaining add or updated the following configuration options in your gdk.yml file:

env:
  DEVELOPMENT_AI_GATEWAY_URL: http://gdk.test:5052
  FETCH_MODEL_SELECTION_DATA_FROM_LOCAL: '1'
  AIGW_AUTH__BYPASS_EXTERNAL: true
gdk:
  license_provisioning:
    duo:
      tier: enterprise
    enabled: true

This should allow you to bypass the auth checks locally

Also ensure to access the GDK via gdk.test:8080 instead of gdk.test:3000 otherwise the AI gateway will error out

4. Create a project inside the Group you enabled Duo for and go to the Pipeline Editor. You should see the new Empty State with the AI card

Note*: If you click the Analyze repository, the chat will default to Duo classic, the Foundational Agent where that this button will make a call for is not yet available

Note**: The AI card won't show up at all if you're running GDK with FOSS_ONLY=1, and this is by design, AI features are EE only

MR acceptance checklist

Evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.

What does this MR do and why?

Create new Empty State for the Pipeline Editor

This empty state reflects uses the new Visual Language that will be included as part of the update_visual_language feature flag

References

• Issue: #590980
• Rollout Issue for the Feature Flag: #590044

Screenshots or screen recordings

Before	After

How to set up and validate locally

Without the AI Card

1. Enable the update_visual_language feature flag

$ echo "Feature.enable(:update_visual_language)" | gdk rails c

2. Go to a project that has no CI configuration
3. Go to Build -> Pipeline Editor

With the AI Card

1. Enable the update_visual_language feature flag

$ echo "Feature.enable(:update_visual_language)" | gdk rails c

2. Follow the steps to enable Duo in your local GDK
3. Go to a group and ensure Duo is enabled for that group, e.g. with in the FlightJS group

Settings -> General -> GitLab Duo Settings:

If you open Duo and you get a message that there are no credits remaining add or updated the following configuration options in your gdk.yml file:

env:
  DEVELOPMENT_AI_GATEWAY_URL: http://gdk.test:5052
  FETCH_MODEL_SELECTION_DATA_FROM_LOCAL: '1'
  AIGW_AUTH__BYPASS_EXTERNAL: true
gdk:
  license_provisioning:
    duo:
      tier: enterprise
    enabled: true

This should allow you to bypass the auth checks locally

Also ensure to access the GDK via gdk.test:8080 instead of gdk.test:3000 otherwise the AI gateway will error out

4. Create a project inside the Group you enabled Duo for and go to the Pipeline Editor. You should see the new Empty State with the AI card

Note*: If you click the Analyze repository, the chat will default to Duo classic, the Foundational Agent where that this button will make a call for is not yet available

Note**: The AI card won't show up at all if you're running GDK with FOSS_ONLY=1, and this is by design, AI features are EE only

MR acceptance checklist

Evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.

Max Fan (b3eeb9de) at 17 Mar 20:15

Max Fan (67ced635) at 17 Mar 19:01

... and 325 more commits

Thanks for this @stomlinson , this was a very good find!

It looks like build was originally passed because there was a FF check: cda4b66e

But when the FF was removed, we never removed the method parameter: 2d3a3323

What does this MR do and why?

Remove unneeded build arg to with_read_consistency

Ci::BuildTraceChunk#with_read_consistency took a build argument, but that argument was unused in the method body. Calling this method was the only reason Ci::BuildTraceChunkFlushWorker was loading a Ci::Build, so removing the unused argument makes the worker cheaper.

This should help reduce replica lock contention happening in gitlab-com/gl-infra/production#21562

References

gitlab-com/gl-infra/production#21562

MR acceptance checklist

Evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.

Looks like the tests are failing because of the extra query count 😆

We should probably just call one or the other here and not both