LUCI Analysis

LUCI Analysis is a system designed to understand and reduce the impact of test failures.

Prerequisites

Commands below assume you are running in the infra environment. To enter the infra env (via the infra.git checkout), run:

eval infra/go/env.py

Local Development flow

To run the server locally, first authorize as the correct GCP project (you should only need to do this once):

gcloud config set project luci-analysis-dev
gcloud auth application-default login

Authenticate in LUCI and in CIPD:

In LUCI Analysis's service-api directory run:

luci-auth login -scopes "https://www.googleapis.com/auth/cloud-platform https://www.googleapis.com/auth/userinfo.email"

In the same directory run:
```
cipd auth-login
```

To run the server, in another terminal use:

cd service-api
go run main.go \
 -cloud-project luci-analysis-dev \
 -spanner-database projects/luci-analysis-dev/instances/dev/databases/luci-analysis-dev \
 -auth-service-host chrome-infra-auth-dev.appspot.com \
 -luci-analysis-host 127.0.0.1:8800 \
 -default-request-timeout 10m0s \
 -buganizer-mode disable \
 -config-local-dir ../configs \
 -encrypted-cookies-expose-state-endpoint

-default-request-timeout is needed if exercising cron jobs through the admin portal as cron jobs run through the /admin/ endpoint attract the default timeout of 1 minute, instead of the 10 minute timeout of the /internal/ endpoint (hit by GAE cron jobs when they are actually executing).

-buganizer-mode is needed when running cron jobs that create/update buganizer issues (e.g. update-analysis-and-bugs). Set the mode to disable to prevent the cron job from filing buganizer bugs.

Note that -config-local-dir is required only if you plan on modifying config and loading it into Cloud Datastore via the read-config cron job accessible via http://127.0.0.1:8900/admin/portal/cron for testing. Omitting this, the server will fetch the current config from Cloud Datastore (as periodically refreshed from LUCI Config Service).

You may also be able to use an arbitrary cloud project (e.g. ‘dev’) if you setup Cloud Datastore emulator and setup a config for that project under configs.

Validating your changes in the LUCI UI

If you need to validate your changes in LUCI UI please refer to Milo's documentation on how to set it up to point to a locally running LUCI Analysis service.

Run Spanner integration tests using Cloud Spanner Emulator

Install Cloud Spanner Emulator

Linux

The Cloud Spanner Emulator is part of the bundled gcloud, to make sure it's installed:

cd infra
gclient runhooks
eval `./go/env.py`
which gcloud # should show bundled gcloud
gcloud components list # should see cloud-spanner-emulator is installed

Run tests

From command line, first set environment variables:

export INTEGRATION_TESTS=1

Then run go test as usual. For example:

go test go.chromium.org/luci/analysis/...

Deployment

Developer Testing

LUCI Analysis uses gae.py for deployment of the GAE instances for developer testing (e.g. of local changes).

First, enter the infra env (via the infra.git checkout):

eval infra/go/env.py

Then use the following commands to deploy:

gae.py upload -A luci-analysis-dev --host-scheme VERSION.staging.analysis.api.luci.app default api

Dev and Prod Instances

The dev and prod instances are managed via LUCI GAE Automatic Deployment (Googlers-only).

Make a production release

Make sure that you have an infra_internal checkout.
Navigate to data/gae directory under your base checkout directory, should be ~/infra.
Pull the latest with git rebase-update or git checkout main && git pull.
Create a new branch git new-branch <NAME> or git -b <NAME>.
run ./scripts/promote.py luci-analysis --canary --stable --commit.
Upload the CL git cl upload.
Request approval from space: LUCI Test War Room.