Lago v2 global - #190
Closed
IxDay wants to merge 86 commits into
Closed
Conversation
Refactor Taskfile to use a map structure for charts with their types (deployment, cronjob, or empty for umbrella/config charts). Add shared test pattern files in tests-common/ that validate extraEnv, extraEnvFrom, and resources configurations. Enable parallel test execution with chart-specific dependencies. - Add extraEnv and extraEnvFrom support to lago-data-worker, lago-data-forecasted-usage, and lago-mcp-server templates - Create test fixtures with required values for each chart - Implement test:unit:* wildcard task for individual chart testing
Add conditional environment variables for WEB_CONCURRENCY, RAILS_MAX_THREADS, and RAILS_MIN_THREADS in the deployment template so they are only set when explicitly configured. Remove the hardcoded default of 4 for concurrency in values.yaml, allowing Puma to use its own defaults when no value is provided.
IxDay
force-pushed
the
lago-v2-global
branch
from
February 10, 2026 13:26
8ae8ceb to
0c458a7
Compare
Bundles a Gotenberg deployment (HTML-to-PDF service) with an optional lago-rails worker subchart for PDF generation jobs. The worker can be disabled via worker.enabled to deploy only Gotenberg.
When the release name is a prefix of the nameOverride (e.g. release "lago" with nameOverride "lago-pdf-worker"), the standard Helm contains check fails because the override is longer than the release name. This caused duplicated prefixes like "lago-pdf-lago-pdf-worker". Add an `else if contains .Release.Name $name` branch so that when the release name is already embedded in the computed name, use the name directly instead of concatenating again. Also updates lago-pdf config nameOverride to lago-pdf-config, reworks the gotenberg fullname helper to build on lago-pdf.fullname, adds fullname unit tests for lago-pdf, and adds lago-pdf to the Taskfile.
Breaking change in fullname helpers (reverse contains logic) affects resource naming for existing deployments, warranting a minor version bump across all charts.
Ensures configs are pushed before components and components before umbrella charts, preventing dependency resolution failures during OCI publishing.
Remove the separate gotenberg.fullname helper in favor of the standard lago-pdf.fullname, reducing unnecessary naming indirection. Switch the Gotenberg image from upstream gotenberg/gotenberg to getlago/lago-gotenberg and add default args to disable health check logging. Update all Chart.lock files to reflect the 0.2.0 dependency versions.
…nit tests Add a `configmap.create` flag to both lago-config and lago-data-config charts, allowing users to skip ConfigMap creation when using an externally managed ConfigMap. This mirrors the existing `secret.create` pattern already present in both charts. Improve values.yaml comments to clearly describe the toggle behavior and document the name resolution fallback paths for each chart. Add helm-unittest coverage for lago-data-config (previously had none): - configmap: create toggle, name resolution, and priority tests - secret: create toggle, name resolution, and priority tests Extend lago-config configmap tests with create toggle assertions.
…rkers Convert all extraEnv fields from list format to deep-mergeable map format across every chart. Update all templates to iterate over map entries instead of using `with`. Add consistent commented examples (simple value, secretKeyRef, configMapKeyRef) to every extraEnv entry in values.yaml. Restructure sidekiq configuration from flat `sidekiqPro` flag to a `sidekiq.pro` and `sidekiq.queues` structure, enabling dedicated worker deployments per queue (analytics, billing, clock, events, webhook, pdf). Add corresponding subchart dependencies and values for each worker. Bump appVersion to v1.41.3 across lago-rails, lago-front, and lago-events-processor-worker. Fix lago-pdf config dependency condition from `worker.enabled` to `config.enabled`. Add global.lago.version field.
Move sidekiqPro from global.lago into a dedicated global.sidekiq section with per-queue toggles (analytics, billing, clock, events, webhook, pdf) to support independent sidekiq queue worker deployments. - Bump appVersion to v1.41.0 across lago-rails, lago-front, and lago-events-processor-worker charts - Add appVersion to umbrella lago chart - Fix front image repository from getlago/app to getlago/front - Switch postgres image to getlago/postgres-partman:15.0-alpine - Add lago.configMapName and lago.secretName helper templates in the umbrella chart and use them consistently in migrate-job - Fix example DNS names to include default namespace - Add basic_workers.yaml example demonstrating sidekiq queue workers - Regenerate Chart.lock files
Add missing CI files for the lago chart e2e test: - postgres.yml and redis.yml manifests for kind cluster infrastructure - e2e-values.yaml with required config for chart-testing install Scope ct install to specific charts in Taskfile, chart-testing-lago, and chart-testing-lago-mcp-server workflows to prevent sub-charts from being installed independently without their infrastructure. Expand chart-testing-lago.yml path triggers to include sub-chart dependencies (lago-config, lago-rails, lago-front, etc.) so the e2e test runs when any dependency changes. Fix redis key names in lago values.yaml (url -> uri) to match what lago-config templates actually reference.
IxDay
force-pushed
the
lago-v2-global
branch
from
February 16, 2026 20:11
e1aab36 to
fe29ec0
Compare
Add missing global keys to the lago umbrella values.yaml: - global.lago.version - global.nango.publicKey and global.nango.secretKey - global.s3.region default to "us-east-1" Rename redis global keys from `url` to `uri` in lago-rails, lago-pdf, and lago-events-processor-worker to match what lago-config templates actually reference.
The gotenberg deployment uses the chart-level lago-pdf.fullname helper directly (resolving to lago-pdf), not a gotenberg-suffixed variant. Update test expectations and override keys accordingly.
- Disable check-version-increment in ct.yaml (v2 rewrite uses 0.2.0 which is lower than the old 1.27.1 on main) - Fix yamllint comment spacing (2 spaces before inline comments) across lago-config, lago-rails, lago-pdf, lago-data, lago-data-worker, lago-data-forecasted-usage, and lago umbrella values.yaml - Fix bracket spacing in lago-data-forecasted-usage - Create missing lago-staging/values.yaml - Add global.dbtPipeline and global.databaseReplica defaults to lago-data-config and lago-data-worker to prevent nil pointer on template rendering - Add config.api.url default to lago-data-worker
Add two companion overlays for basic_workers.yaml: - basic_workers_resources.yaml: production resource requests/limits and HPA autoscaling with minReplicas: 1 for all components. Uses no CPU limits (avoids throttling) and memory request = limit (guaranteed QoS). - basic_workers_scaled.yaml: minReplicas overrides matching recommended production replica counts. Layers on top of the resources overlay. Usage combinations: basic_workers.yaml alone = staging (no resources, no HPA) + resources overlay = production-like (resources + HPA starting at 1) + resources + scaled = full production (recommended min replicas)
Move ct.yaml, cr.yaml, chart_schema.yaml, and lintconf.yaml into a dedicated tools/ directory to declutter the repo root. Update all references in Taskfile.yaml and GitHub Actions workflows. Add chart-yaml-schema and lint-conf paths to ct.yaml so ct can locate its companion files, enabling local linting without the CI Docker image.
Add docs and docs:* Taskfile targets for generating chart README files using helm-docs. The docs:* target allows generating docs for a single chart (e.g. task docs:lago-rails), while docs generates for all charts. Generate initial README.md for all 13 charts from their existing Chart.yaml metadata and values.yaml comments.
Add `# --` description comments to every field in all 13 chart values.yaml files for helm-docs generation. Covers global shared values, subchart overrides, container/image/service config, autoscaling, ingress, probes, and all data/streaming settings. Regenerate all chart READMEs with the new descriptions.
- Add @section annotations to all values.yaml files across all 13 charts to group parameters into logical sections (Lago, Config, Database, etc.) - Consolidate Scheduling, Volumes, Probes, Resources, and Environment sections into the Pod section for cleaner README organization - Replace long inline JSON blobs for complex object defaults with @default links to specific lines in values.yaml (probes, httpRoute, ingress.hosts, config) - Add @ignore to parent global: keys to prevent duplicate rendering of the entire global object as a JSON blob in "Other Values" section - Regenerate all chart READMEs with helm-docs
Replace the old monolithic parameter tables with a streamlined README that points to the lago umbrella chart as the main entry point. Link to the example overlays (basic, basic_workers, resources, scaled) for getting started, document the kind-based local dev workflow, and reference each chart's auto-generated README for detailed values.
Bump chart version and dependency versions from 0.2.0 to 0.3.0 across all 13 charts. Regenerate READMEs to reflect the new version badges.
Add mise.toml with pinned versions for helm, helm-docs, helm-unittest, chart-testing, kind, kubectl, and yamllint. Replace `helm unittest` calls in Taskfile with `untt`, the binary name installed by mise.
Add @default annotations to parent keys in lago and lago-data umbrella charts so helm-docs renders "See child values" instead of serializing the entire object tree as an inline JSON blob. Applies to 16 parent keys in lago (api, worker, clock, pdf, migrate, etc.) and 6 in lago-data.
The lago-data-config subchart provides helper templates (configMapName, secretName) that are always needed by the data-worker. With config.enabled set to false, Helm skips loading the subchart entirely in CI (clean checkout with no unpacked chart directory), causing template resolution failures. Locally it worked because the gitignored unpacked subchart directory persisted from previous builds. Default to true for standalone deployments; the lago-data umbrella chart already overrides this to false when providing config at the parent level.
Move Redis and Postgres manifests from standalone CI files into extraObjects within e2e-values.yaml and basic.yaml. This eliminates the need for separate kubectl apply steps before chart-testing, letting helm hooks manage dependency ordering automatically. - Delete charts/lago/ci/postgres.yml and redis.yml - Remove kubectl apply steps from Taskfile e2e task - Add pre-install helm hook annotations to all extraObjects - Add liveness probes to Redis and Postgres containers - Add oci:install Taskfile task for installing from GHCR
Helm does not wait for Deployment hooks to be ready before proceeding to the next hook weight. This means the migrate job (weight -5) starts before Redis/Postgres (weight -10) pods are actually accepting connections. Add initContainers support to the migrate job template and configure wait-for-db and wait-for-redis init containers in all examples that deploy postgres/redis via extraObjects. Also add busybox image to kind image preloading.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.