Conversation
remove docs db schema for users
- Adds docs for analytics + errors - Rewrite of selects and others - Tranforming ch results - Other stuff im missing
|
|
The latest updates on your projects. Learn more about Vercel for GitHub.
1 Skipped Deployment
|
📝 WalkthroughWalkthroughAdds a new analytics feature: POST /v2/analytics.getVerifications endpoint with SQL parser/enforcement, per-workspace ClickHouse connection manager and provisioning, RBAC Changes
Sequence Diagram(s)sequenceDiagram
participant Client
participant APIServer as API Server
participant Keys as Key Service
participant Parser as SQL Parser
participant ConnMgr as Analytics ConnMgr
participant CH as ClickHouse
Client->>APIServer: POST /v2/analytics.getVerifications { query }
APIServer->>Keys: validate root key, fetch permissions
Keys-->>APIServer: permissions (api.*.read_analytics / api.<id>.read_analytics)
APIServer->>Parser: Parse & rewrite query (SELECT-only, functions, tables)
alt parser error
Parser-->>APIServer: error (400)
APIServer-->>Client: 400 Bad Request
else
Parser-->>APIServer: rewritten SQL + injected security filters
APIServer->>ConnMgr: GetConnection(workspaceID)
alt not configured / conn fail
ConnMgr-->>APIServer: 412/503
APIServer-->>Client: 412/503
else
ConnMgr-->>APIServer: ClickHouse client
APIServer->>CH: Execute rewritten SQL
alt resource/quota error
CH-->>APIServer: 422/429
APIServer-->>Client: 422/429
else
CH-->>APIServer: rows ([]map[string]any)
APIServer-->>Client: 200 { meta, data: rows }
end
end
end
Estimated code review effort🎯 5 (Critical) | ⏱️ ~120 minutes Areas needing extra attention:
Possibly related PRs
Suggested labels
Pre-merge checks and finishing touches❌ Failed checks (1 warning, 1 inconclusive)
✅ Passed checks (2 passed)
✨ Finishing touches🧪 Generate unit tests (beta)
📜 Recent review detailsConfiguration used: CodeRabbit UI Review profile: CHILL Plan: Pro 📒 Files selected for processing (4)
🚧 Files skipped from review as they are similar to previous changes (2)
🧰 Additional context used🧠 Learnings (5)📓 Common learnings📚 Learning: 2025-02-24T13:35:00.816ZApplied to files:
📚 Learning: 2025-08-20T11:41:36.718ZApplied to files:
📚 Learning: 2025-08-17T22:54:06.787ZApplied to files:
📚 Learning: 2025-09-24T18:57:34.843ZApplied to files:
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (6)
🔇 Additional comments (7)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
coderabbitai
bot
requested review from
MichaelUnkey,
imeyer,
mcstepp and
perkinsjr
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
go/apps/api/routes/v2_analytics_get_verifications/422_test.go (1)
63-68: Avoid fixed 10 s sleep in this test.A hard sleep adds 10 s to every run and still risks flaking if buffering ever takes longer. Please poll until the condition is met (or times out) so the test fails fast when ready.
- // Wait for data to be buffered and flushed to ClickHouse - time.Sleep(10 * time.Second) - - // Query should fail with 422 due to max_memory_usage limit being exceeded - res := testutil.CallRoute[Request, Response](h, route, headers, req) - require.Equal(t, 422, res.Status) + var res testutil.TestResponse[Response] + require.Eventually(t, func() bool { + res = testutil.CallRoute[Request, Response](h, route, headers, req) + return res.Status == http.StatusUnprocessableEntity + }, 30*time.Second, 500*time.Millisecond)
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (16)
apps/engineering/content/docs/architecture/services/analytics.mdx(1 hunks)go/apps/api/routes/v2_analytics_get_verifications/422_test.go(1 hunks)go/apps/api/routes/v2_analytics_get_verifications/503_test.go(1 hunks)go/cmd/create-clickhouse-user/main.go(1 hunks)go/pkg/clickhouse/user.go(1 hunks)go/pkg/db/bulk_clickhouse_workspace_settings_insert.sql_generated.go(1 hunks)go/pkg/db/clickhouse_workspace_settings_find_by_workspace_id.sql_generated.go(1 hunks)go/pkg/db/clickhouse_workspace_settings_insert.sql_generated.go(1 hunks)go/pkg/db/clickhouse_workspace_settings_update_limits.sql_generated.go(1 hunks)go/pkg/db/models_generated.go(1 hunks)go/pkg/db/querier_generated.go(7 hunks)go/pkg/db/queries/clickhouse_workspace_settings_insert.sql(1 hunks)go/pkg/db/queries/clickhouse_workspace_settings_update_limits.sql(1 hunks)go/pkg/db/schema.sql(1 hunks)go/pkg/testutil/http.go(4 hunks)internal/db/src/schema/clickhouse_workspace_settings.ts(1 hunks)
✅ Files skipped from review due to trivial changes (1)
- apps/engineering/content/docs/architecture/services/analytics.mdx
🚧 Files skipped from review as they are similar to previous changes (7)
- go/pkg/db/models_generated.go
- go/apps/api/routes/v2_analytics_get_verifications/503_test.go
- go/pkg/db/queries/clickhouse_workspace_settings_update_limits.sql
- go/pkg/db/queries/clickhouse_workspace_settings_insert.sql
- go/pkg/db/bulk_clickhouse_workspace_settings_insert.sql_generated.go
- go/pkg/db/clickhouse_workspace_settings_find_by_workspace_id.sql_generated.go
- go/pkg/db/clickhouse_workspace_settings_insert.sql_generated.go
🧰 Additional context used
🧠 Learnings (12)
📓 Common learnings
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 4190
File: go/internal/services/keys/verifier.go:51-53
Timestamp: 2025-10-30T15:10:52.743Z
Learning: PR #4190 for unkeyed/unkey is focused solely on database schema and query changes for identity-based credits. It adds IdentityCredits and KeyCredits fields to structs and queries, but does not implement the priority enforcement logic in the usagelimiter. The logic implementation is intentionally deferred to a later PR in the stack.
📚 Learning: 2025-04-22T14:40:51.459Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3161
File: go/pkg/clickhouse/schema/databases/001_verifications/002_raw_key_verifications_v1.sql:31-33
Timestamp: 2025-04-22T14:40:51.459Z
Learning: The ClickHouse table schemas in the codebase mirror the production environment and cannot be modified directly in PRs without careful migration planning.
Applied to files:
internal/db/src/schema/clickhouse_workspace_settings.tsgo/pkg/db/schema.sql
📚 Learning: 2024-10-04T20:44:38.489Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 2180
File: apps/dashboard/lib/constants/workspace-navigations.tsx:56-118
Timestamp: 2024-10-04T20:44:38.489Z
Learning: When typing the `workspace` parameter in functions like `createWorkspaceNavigation`, prefer importing the `Workspace` type from the database module and picking the necessary keys (e.g., `features`) instead of redefining the interface.
Applied to files:
internal/db/src/schema/clickhouse_workspace_settings.ts
📚 Learning: 2025-09-12T08:01:20.792Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3944
File: go/pkg/db/acme_challenge_update_verified_with_expiry.sql_generated.go:31-39
Timestamp: 2025-09-12T08:01:20.792Z
Learning: Do not review or suggest changes to files with sql_generated.go suffix or other files marked as auto-generated (containing "Code generated by" comments), as these are generated by tools like sqlc and changes would be overwritten on regeneration.
Applied to files:
go/pkg/db/clickhouse_workspace_settings_update_limits.sql_generated.go
📚 Learning: 2025-07-16T15:38:53.491Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3606
File: go/pkg/db/replica.go:8-11
Timestamp: 2025-07-16T15:38:53.491Z
Learning: For debugging database replica usage in go/pkg/db/replica.go, it's acceptable to mark QueryRowContext operations as "success" even though SQL errors only surface during row.Scan() calls. The timing metrics are the primary concern for debugging replica performance patterns.
Applied to files:
go/pkg/testutil/http.go
📚 Learning: 2025-08-14T16:25:48.167Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3785
File: go/apps/api/routes/v2_keys_reroll_key/401_test.go:52-61
Timestamp: 2025-08-14T16:25:48.167Z
Learning: User Flo4604 requested creation of a GitHub issue to track converting all test files to use table-driven test patterns as a broader codebase improvement, following the suggestion made during review of go/apps/api/routes/v2_keys_reroll_key/401_test.go.
Applied to files:
go/apps/api/routes/v2_analytics_get_verifications/422_test.go
📚 Learning: 2025-04-08T09:34:24.576Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 2872
File: apps/dashboard/lib/trpc/routers/ratelimit/createNamespace.ts:36-39
Timestamp: 2025-04-08T09:34:24.576Z
Learning: In the Unkey dashboard, when making database queries involving workspaces, use `ctx.workspace.id` directly instead of fetching the workspace separately for better performance and security.
Applied to files:
go/pkg/db/querier_generated.go
📚 Learning: 2025-01-31T13:50:45.004Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 2825
File: apps/dashboard/app/(app)/logs/components/table/logs-table.tsx:100-118
Timestamp: 2025-01-31T13:50:45.004Z
Learning: Sensitive data in logs (passwords, tokens, etc.) is masked before being written to Clickhouse, ensuring secure storage of log data.
Applied to files:
go/cmd/create-clickhouse-user/main.go
📚 Learning: 2025-02-27T14:35:15.251Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 2918
File: apps/dashboard/lib/trpc/routers/api/overview-api-search.ts:0-0
Timestamp: 2025-02-27T14:35:15.251Z
Learning: When using Drizzle ORM with LIKE queries, use the sql`` tagged template syntax (e.g., sql`${table.name} LIKE ${`%${input.query}%`}`) instead of direct string interpolation to prevent SQL injection attacks.
Applied to files:
go/cmd/create-clickhouse-user/main.go
📚 Learning: 2025-09-15T17:40:51.536Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 3973
File: go/pkg/db/schema.sql:307-308
Timestamp: 2025-09-15T17:40:51.536Z
Learning: The environments table in the Unkey codebase is not being used in production yet, so schema changes to it don't require complex migration sequences to handle existing data or concurrent usage.
Applied to files:
go/cmd/create-clickhouse-user/main.gogo/pkg/db/schema.sql
📚 Learning: 2025-04-22T14:43:11.724Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3161
File: go/pkg/clickhouse/schema/databases/002_ratelimits/006_ratelimits_per_day_v1.sql:1-13
Timestamp: 2025-04-22T14:43:11.724Z
Learning: In the unkey project, the SQL files in clickhouse/schema/databases represent the current production schema and shouldn't be modified directly in PRs. Schema changes require dedicated migration scripts.
Applied to files:
go/pkg/db/schema.sql
📚 Learning: 2025-10-21T09:45:47.560Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 4107
File: go/pkg/clickhouse/key_verifications_test.go:20-20
Timestamp: 2025-10-21T09:45:47.560Z
Learning: In ClickHouse tests (e.g., go/pkg/clickhouse/key_verifications_test.go), parallel execution with t.Parallel() is safe when tests use workspaceID-based isolation. Each test generates a unique workspaceID (via uid.New(uid.WorkspacePrefix)), and all database operations are scoped to that workspaceID, providing logical isolation even when tests share a single ClickHouse instance.
Applied to files:
go/pkg/db/schema.sql
🧬 Code graph analysis (5)
internal/db/src/schema/clickhouse_workspace_settings.ts (1)
internal/db/src/schema/workspaces.ts (1)
workspaces(17-105)
go/pkg/testutil/http.go (8)
go/internal/services/caches/caches.go (2)
Caches(18-42)New(145-266)go/pkg/testutil/containers/containers.go (1)
ClickHouse(130-132)go/internal/services/analytics/connection_manager.go (3)
ConnectionManager(22-24)NewConnectionManager(47-85)ConnectionManagerConfig(37-44)go/pkg/testutil/seed/seed.go (2)
Seeder(31-36)New(39-46)go/gen/proto/vault/v1/service.pb.go (3)
EncryptRequest(104-110)EncryptRequest(123-123)EncryptRequest(138-140)go/pkg/clickhouse/user.go (2)
UserConfig(11-28)DefaultAllowedTables(143-152)go/pkg/db/clickhouse_workspace_settings_insert.sql_generated.go (1)
InsertClickhouseWorkspaceSettingsParams(42-54)go/gen/proto/vault/v1/object.pb.go (3)
Encrypted(252-264)Encrypted(277-277)Encrypted(292-294)
go/apps/api/routes/v2_analytics_get_verifications/422_test.go (5)
go/pkg/testutil/http.go (3)
NewHarness(59-205)WithMaxQueryMemoryBytes(270-274)CallRoute(383-417)go/pkg/testutil/seed/seed.go (1)
CreateApiRequest(84-92)go/pkg/clickhouse/schema/requests.go (1)
KeyVerificationRequestV1(66-92)go/pkg/uid/uid.go (2)
RequestPrefix(19-19)KeyPrefix(16-16)go/apps/api/routes/v2_analytics_get_verifications/handler.go (3)
Handler(49-56)Request(26-26)Response(27-27)
go/pkg/db/querier_generated.go (9)
go/pkg/partition/db/database.go (1)
DBTX(10-10)go/pkg/db/models_generated.go (2)
ClickhouseWorkspaceSetting(569-581)Identity(643-652)go/pkg/db/identity_find_many.sql_generated.go (1)
FindIdentitiesParams(21-25)go/pkg/db/api_find_key_auth_by_ids.sql_generated.go (2)
FindKeyAuthsByIdsParams(23-26)FindKeyAuthsByIdsRow(28-31)go/pkg/db/api_find_key_auth_by_key_auth_ids.sql_generated.go (2)
FindKeyAuthsByKeyAuthIdsParams(23-26)FindKeyAuthsByKeyAuthIdsRow(28-31)go/pkg/db/key_auth_get_by_id.sql_generated.go (1)
GetKeyAuthByIDRow(26-33)go/pkg/db/clickhouse_workspace_settings_insert.sql_generated.go (1)
InsertClickhouseWorkspaceSettingsParams(42-54)go/pkg/db/key_auth_insert.sql_generated.go (1)
InsertKeyAuthParams(31-37)go/pkg/db/clickhouse_workspace_settings_update_limits.sql_generated.go (1)
UpdateClickhouseWorkspaceSettingsLimitsParams(26-35)
go/cmd/create-clickhouse-user/main.go (8)
go/pkg/cli/flag.go (6)
Flag(20-27)String(419-451)Required(298-317)StringSlice(612-644)Int(532-569)Int64(647-684)go/pkg/cli/command.go (1)
Action(21-21)go/pkg/clickhouse/client.go (1)
Config(37-44)go/apps/api/config.go (2)
Config(23-105)S3Config(16-21)go/pkg/vault/storage/interface.go (1)
Storage(15-30)go/pkg/db/clickhouse_workspace_settings_insert.sql_generated.go (1)
InsertClickhouseWorkspaceSettingsParams(42-54)go/pkg/db/clickhouse_workspace_settings_update_limits.sql_generated.go (1)
UpdateClickhouseWorkspaceSettingsLimitsParams(26-35)go/pkg/clickhouse/user.go (1)
UserConfig(11-28)
🔇 Additional comments (1)
go/pkg/db/querier_generated.go (1)
151-155: LGTM! Generated code aligns with PR objectives.The newly added interface methods support the analytics feature implementation:
- ClickHouse workspace settings management (
FindClickhouseWorkspaceSettingsByWorkspaceID,InsertClickhouseWorkspaceSettings,UpdateClickhouseWorkspaceSettingsLimits)- Identity batch operations (
FindIdentities)- Key auth lookups (
FindKeyAuthsByIds,FindKeyAuthsByKeyAuthIds,GetKeyAuthByID,InsertKeyAuth)All methods follow consistent naming conventions and parameter patterns. Since this is sqlc-generated code, any necessary adjustments should be made to the source SQL queries rather than this file.
Also applies to: 273-280, 344-363, 872-884, 1003-1031, 1199-1216, 1818-1830
There was a problem hiding this comment.
Actionable comments posted: 7
♻️ Duplicate comments (5)
go/Makefile (1)
37-38: Pin the buf CLI version ingo run(duplicate of past concern).Both lines now invoke
go run github.com/bufbuild/buf/cmd/bufwithout an@versionspecifier, which causes non-reproducible fetches and pollutesgo.sum. This reintroduces the issue flagged in the previous review.Please pin a specific buf release (e.g.,
@vX.Y.Z) or use aBUF_VERSIONMake variable for both invocations somake generateremains deterministic.Apply this diff to pin the buf version:
- go run github.com/bufbuild/buf/cmd/buf generate --template ./buf.gen.connect.yaml --clean --path "./proto/cache" --path "./proto/ctrl" --path "./proto/krane" --path "./proto/partition" --path "./proto/vault" - go run github.com/bufbuild/buf/cmd/buf generate --template ./buf.gen.restate.yaml --path "./proto/hydra" + go run github.com/bufbuild/buf/cmd/buf@v1.32.2 generate --template ./buf.gen.connect.yaml --clean --path "./proto/cache" --path "./proto/ctrl" --path "./proto/krane" --path "./proto/partition" --path "./proto/vault" + go run github.com/bufbuild/buf/cmd/buf@v1.32.2 generate --template ./buf.gen.restate.yaml --path "./proto/hydra"(Replace
v1.32.2with the appropriate buf version your project uses.)go/pkg/codes/generate.go (2)
253-264: URN parsing drops every error code.The URN parsing assumes
parts[0] == "err"but real URNs start withurn:err:..., causing all error codes to be skipped. No MDX files will be generated.Apply this fix to handle both formats:
// Parse URN to get file path - // Example: err:user:bad_request:client_closed_request -> user/bad_request/client_closed_request.mdx + // Example: urn:err:user:bad_request:client_closed_request -> user/bad_request/client_closed_request.mdx parts := strings.Split(errCode.URN, ":") - if len(parts) < 4 || parts[0] != "err" { + + // Handle both "err:..." and "urn:err:..." formats + startIdx := 0 + if len(parts) > 0 && parts[0] == "urn" { + startIdx = 1 + } + if len(parts) < startIdx+4 || parts[startIdx] != "err" { continue } // Build file path from URN parts (skip "err:" prefix) - pathParts := parts[1 : len(parts)-1] + pathParts := parts[startIdx+1 : len(parts)-1] fileName := parts[len(parts)-1] + ".mdx" filePath := filepath.Join(append([]string{baseDocsPath}, append(pathParts, fileName)...)...)
317-326: URN parsing drops every error code (same bug).Same parsing bug as in generateMissingMDXFiles - assumes
parts[0] == "err"when URNs start withurn:err:.... The validPaths map will be incomplete, preventing obsolete file detection.Apply the same fix as above:
parts := strings.Split(errCode.URN, ":") - if len(parts) < 4 || parts[0] != "err" { + + // Handle both "err:..." and "urn:err:..." formats + startIdx := 0 + if len(parts) > 0 && parts[0] == "urn" { + startIdx = 1 + } + if len(parts) < startIdx+4 || parts[startIdx] != "err" { continue } // Build file path from URN - pathParts := parts[1 : len(parts)-1] + pathParts := parts[startIdx+1 : len(parts)-1] fileName := parts[len(parts)-1] + ".mdx" filePath := filepath.Join(append([]string{baseDocsPath}, append(pathParts, fileName)...)...)deployment/docker-compose.yaml (1)
114-125: Critical: Fix Kafka advertised listeners for Docker network.The Kafka service upgrade is correct, but Line 121 configures
KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://127.0.0.1:9092, which advertises the loopback address. This prevents other containers (like apiv2 at line 94) from connecting to Kafka within the Docker network. This issue was previously flagged but remains unresolved.Apply this diff:
- KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://127.0.0.1:9092 + KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://kafka:9092go/apps/api/openapi/openapi-generated.yaml (1)
2029-2067: Fix unsatisfiable allOf: BaseError blocks BadRequestErrorDetails.errorsBadRequestErrorDetails extends BaseError via allOf but BaseError has additionalProperties: false, so the added required property errors is currently disallowed, making the schema impossible to satisfy. Quick fix: declare errors (optional) on BaseError, and keep it required only for BadRequestErrorDetails.
Apply this minimal diff:
components: schemas: BaseError: properties: detail: description: A human-readable explanation specific to this occurrence of the problem. This provides detailed information about what went wrong and potential remediation steps. The message is intended to be helpful for developers troubleshooting the issue. example: Property foo is required but is missing. type: string status: description: HTTP status code that corresponds to this error. This will match the status code in the HTTP response. Common codes include `400` (Bad Request), `401` (Unauthorized), `403` (Forbidden), `404` (Not Found), `409` (Conflict), and `500` (Internal Server Error). example: 404 format: int type: integer title: description: A short, human-readable summary of the problem type. This remains constant from occurrence to occurrence of the same problem and should be used for programmatic handling. example: Not Found type: string type: description: A URI reference that identifies the problem type. This provides a stable identifier for the error that can be used for documentation lookups and programmatic error handling. When followed, this URI should provide human-readable documentation for the problem type. example: https://unkey.com/docs/errors/unkey/resource/not_found type: string + errors: + description: List of individual validation errors that occurred in the request (present on 400 responses). + type: array + items: + $ref: "#/components/schemas/ValidationError" required: - title - detail - status - type type: object additionalProperties: false
🧹 Nitpick comments (7)
go/pkg/codes/generate.go (2)
238-303: Extract duplicated URN parsing logic.The URN parsing logic (splitting, validating, extracting path parts) is duplicated across all three functions. Consider extracting it into a helper function to improve maintainability.
Example helper:
// parseErrorURN parses an error URN and returns the path parts and filename, // or an error if the URN format is invalid. func parseErrorURN(urn string) (pathParts []string, fileName string, err error) { parts := strings.Split(urn, ":") // Handle both "err:..." and "urn:err:..." formats startIdx := 0 if len(parts) > 0 && parts[0] == "urn" { startIdx = 1 } if len(parts) < startIdx+4 || parts[startIdx] != "err" { return nil, "", fmt.Errorf("invalid URN format: %s", urn) } pathParts = parts[startIdx+1 : len(parts)-1] fileName = parts[len(parts)-1] return pathParts, fileName, nil }Also applies to: 305-369, 371-576
401-418: Type assertions without safety checks.Lines 401, 412, and 418 perform type assertions without checking the
okvalue, which could panic if the docs.json structure is unexpected. While acceptable for a build-time tool (panics help catch structure changes early), adding safety checks would make debugging easier.Example for line 401:
- dd := dropdown.(map[string]interface{}) + dd, ok := dropdown.(map[string]interface{}) + if !ok { + continue + }go/pkg/zen/middleware_errors.go (1)
304-316: Verify response type naming.The analytics connection failure correctly returns 503 status, but it uses
InternalServerErrorResponseas the response type. While this works functionally (since thestatusfield is set to 503), the type name suggests it's for 500 errors. Consider whether a dedicatedServiceUnavailableErrorResponsetype should be added to the OpenAPI schema for semantic clarity, or verify this reuse is intentional.go/apps/api/openapi/openapi-generated.yaml (3)
97-108: Constrain query input and update example to current table aliasAdd sane length bounds to reduce abuse and align with backend limits; update example to the public alias the parser supports (no version suffix).
V2AnalyticsGetVerificationsRequestBody: type: object required: - query properties: query: type: string description: | SQL query to execute against your analytics data. Only SELECT queries are allowed. - example: "SELECT COUNT(*) as total FROM key_verifications_v1 WHERE outcome = 'VALID' AND time >= now() - INTERVAL 7 DAY" + minLength: 7 + maxLength: 20000 + example: "SELECT COUNT(*) AS total FROM key_verifications WHERE outcome = 'VALID' AND time >= now() - INTERVAL 7 DAY"Please confirm the canonical user-facing table/alias (e.g., key_verifications) so the example matches the parser’s rewrite rules. Based on learnings.
3331-3399: Document required permissions and confirm endpoint visibilityThis endpoint omits a “Required Permissions” section while other endpoints include it. Add explicit RBAC docs for api.*.read_analytics and api.<api_id>.read_analytics. Also, it's marked x-hidden and x-speakeasy-ignore — verify this is intentional for the published bundle.
/v2/analytics.getVerifications: post: description: | Execute custom SQL queries against your key verification analytics. For complete documentation including available tables, columns, data types, query examples, see the schema reference in the API documentation. + + Required Permissions + + Your root key must have one of: + - `api.*.read_analytics` + - `api.<api_id>.read_analytics` @@ - x-hidden: true - x-speakeasy-ignore: true + # TODO: Set to false/omit when ready to publish in docs/SDKs + x-hidden: true + x-speakeasy-ignore: trueIf the endpoint should be public at launch, drop the hidden flags in the generated bundle/template.
2094-2108: Optional: tighten result row schema for better DXYou can keep items as open objects but hint common fields to improve SDK typings (non-breaking).
V2AnalyticsGetVerificationsResponseData: type: array description: Array of verification rows returned by the query. Fields vary based on the SQL SELECT clause. items: - type: object - additionalProperties: true - description: Dynamic row with fields determined by the query. Can include any combination of fields like time, outcome, count, key_id, etc. + type: object + additionalProperties: true + properties: + time: + type: integer + format: int64 + description: Unix ms timestamp (when selected) + outcome: + type: string + description: Verification outcome (when selected) + count: + type: integer + format: int64 + description: Aggregated row count (when selected) + key_id: + type: string + description: Key identifier (when selected) + description: Dynamic row with fields determined by the query; common fields are hinted for tooling only.go/main.go (1)
8-11: Optional: gate dev-only commands behind build tagsIf you want to exclude dev commands from release builds, split files and use tags.
Example:
- dev "github.com/unkeyed/unkey/go/cmd/dev" +//go:build dev +// +build dev +dev "github.com/unkeyed/unkey/go/cmd/dev"And wrap the Commands entry in a dev-tagged file:
//go:build dev package main func init() { app.Commands = append(app.Commands, dev.Cmd) }
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
⛔ Files ignored due to path filters (1)
go/go.sumis excluded by!**/*.sum
📒 Files selected for processing (25)
apps/docs/analytics/getting-started.mdx(1 hunks)apps/docs/docs.json(8 hunks)apps/docs/errors/user/bad_request/invalid_analytics_query.mdx(1 hunks)apps/docs/errors/user/unprocessable_entity/query_rows_limit_exceeded.mdx(1 hunks)deployment/docker-compose.yaml(3 hunks)go/Makefile(2 hunks)go/apps/api/config.go(1 hunks)go/apps/api/openapi/gen.go(4 hunks)go/apps/api/openapi/openapi-generated.yaml(6 hunks)go/apps/api/openapi/spec/paths/v2/analytics/getVerifications/index.yaml(1 hunks)go/apps/api/routes/v2_analytics_get_verifications/handler.go(1 hunks)go/apps/api/run.go(2 hunks)go/cmd/api/main.go(2 hunks)go/go.mod(6 hunks)go/internal/services/caches/caches.go(2 hunks)go/main.go(2 hunks)go/pkg/cache/cache.go(4 hunks)go/pkg/cache/clustering/cluster_cache.go(3 hunks)go/pkg/cache/interface.go(1 hunks)go/pkg/cache/middleware/tracing.go(4 hunks)go/pkg/clickhouse/key_verifications_test.go(3 hunks)go/pkg/clickhouse/query-parser/cte.go(1 hunks)go/pkg/codes/generate.go(7 hunks)go/pkg/testutil/http.go(4 hunks)go/pkg/zen/middleware_errors.go(5 hunks)
🚧 Files skipped from review as they are similar to previous changes (6)
- go/cmd/api/main.go
- apps/docs/errors/user/unprocessable_entity/query_rows_limit_exceeded.mdx
- apps/docs/analytics/getting-started.mdx
- go/apps/api/config.go
- apps/docs/errors/user/bad_request/invalid_analytics_query.mdx
- go/apps/api/openapi/spec/paths/v2/analytics/getVerifications/index.yaml
🧰 Additional context used
🧠 Learnings (25)
📓 Common learnings
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 4190
File: go/internal/services/keys/verifier.go:51-53
Timestamp: 2025-10-30T15:10:52.743Z
Learning: PR #4190 for unkeyed/unkey is focused solely on database schema and query changes for identity-based credits. It adds IdentityCredits and KeyCredits fields to structs and queries, but does not implement the priority enforcement logic in the usagelimiter. The logic implementation is intentionally deferred to a later PR in the stack.
📚 Learning: 2025-09-15T19:53:28.487Z
Learnt from: mcstepp
Repo: unkeyed/unkey PR: 3952
File: go/proto/ctrl/v1/routing.proto:0-0
Timestamp: 2025-09-15T19:53:28.487Z
Learning: In the Unkey codebase, authentication/authorization errors intentionally return NOT_FOUND error codes instead of distinct auth error codes (like FORBIDDEN or UNAUTHORIZED) for security reasons. This prevents attackers from distinguishing between "resource doesn't exist" and "you don't have permission to access this resource", avoiding information disclosure about workspace existence and access boundaries.
Applied to files:
go/pkg/zen/middleware_errors.go
📚 Learning: 2025-07-15T14:25:05.608Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3560
File: go/apps/api/routes/v2_keys_create_key/handler.go:353-466
Timestamp: 2025-07-15T14:25:05.608Z
Learning: In the Unkey codebase, input validation for API endpoints is handled at the OpenAPI schema layer, which validates request fields like permission slugs (pattern: "^[a-zA-Z0-9_]+$", length: 1-100 characters) before requests reach the handler code. This validation occurs during the zen.BindBody call in handlers.
Applied to files:
go/pkg/zen/middleware_errors.go
📚 Learning: 2025-03-19T09:25:59.751Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 2955
File: go/apps/api/routes/v2_identities_create_identity/handler.go:162-202
Timestamp: 2025-03-19T09:25:59.751Z
Learning: In the Unkey codebase, input validation for API endpoints is primarily handled through OpenAPI schema validation, which occurs before requests reach the handler code. For example, in the identities.createIdentity endpoint, minimum values for ratelimit duration and limit are defined in the OpenAPI schema rather than duplicating these checks in the handler.
Applied to files:
go/pkg/zen/middleware_errors.go
📚 Learning: 2025-08-08T15:09:01.312Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3753
File: go/apps/api/openapi/config.yaml:9-10
Timestamp: 2025-08-08T15:09:01.312Z
Learning: Repo unkeyed/unkey: For go/apps/api/openapi, oapi-codegen is used and does not support OpenAPI 3.1 union types like [T, "null"]; an overlay step is required to downconvert to 3.0-style nullable before code generation.
Applied to files:
go/apps/api/openapi/gen.gogo/go.mod
📚 Learning: 2025-08-08T15:09:01.312Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3753
File: go/apps/api/openapi/config.yaml:9-10
Timestamp: 2025-08-08T15:09:01.312Z
Learning: Repo unkeyed/unkey: In go/apps/api/openapi, oapi-codegen doesn’t support OpenAPI 3.1 union nullability; overlay.yaml must be applied before codegen. The overlay key in oapi-codegen config isn’t supported—use a pre-step (programmatic or CLI) to merge overlay into the bundled spec, then run oapi-codegen.
Applied to files:
go/apps/api/openapi/gen.go
📚 Learning: 2025-08-21T15:54:45.198Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3825
File: go/internal/services/usagelimiter/limit.go:38-0
Timestamp: 2025-08-21T15:54:45.198Z
Learning: In go/internal/services/usagelimiter/limit.go, the UpdateKeyCreditsDecrement operation cannot be safely wrapped with db.WithRetry due to the lack of idempotency mechanisms in the current tech stack. Retrying this non-idempotent write operation risks double-charging users if the first attempt commits but the client sees a transient error.
Applied to files:
go/apps/api/openapi/gen.go
📚 Learning: 2025-07-16T15:38:53.491Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3606
File: go/pkg/db/replica.go:8-11
Timestamp: 2025-07-16T15:38:53.491Z
Learning: For debugging database replica usage in go/pkg/db/replica.go, it's acceptable to mark QueryRowContext operations as "success" even though SQL errors only surface during row.Scan() calls. The timing metrics are the primary concern for debugging replica performance patterns.
Applied to files:
go/pkg/testutil/http.go
📚 Learning: 2025-07-22T09:02:12.495Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3638
File: deployment/docker-compose.yaml:81-94
Timestamp: 2025-07-22T09:02:12.495Z
Learning: The docker-compose.yaml file in deployment/ is specifically for development environments, not production. Kafka and other service configurations in this file should be optimized for development convenience rather than production security/hardening.
Applied to files:
deployment/docker-compose.yamlgo/Makefile
📚 Learning: 2025-09-01T02:33:43.791Z
Learnt from: imeyer
Repo: unkeyed/unkey PR: 3899
File: go/buf.gen.yaml:0-0
Timestamp: 2025-09-01T02:33:43.791Z
Learning: In the unkeyed/unkey repository, buf commands are executed from the `go/` directory where the `buf.yaml` file is located. This means the `out: gen` configuration in `go/buf.gen.yaml` generates files to `go/gen/` relative to the repository root, which aligns with the expected directory structure.
Applied to files:
go/Makefilego/go.mod
📚 Learning: 2025-09-01T01:57:42.227Z
Learnt from: imeyer
Repo: unkeyed/unkey PR: 3899
File: go/proto/metald/v1/metald.proto:5-9
Timestamp: 2025-09-01T01:57:42.227Z
Learning: In the unkeyed/unkey repository, buf is configured to properly resolve metald proto imports like "metald/v1/vm.proto" without needing the full "go/proto/" prefix. The buf lint command `buf lint --path proto/metald` passes successfully with these relative import paths.
Applied to files:
go/Makefilego/go.mod
📚 Learning: 2025-10-15T10:12:40.810Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 4098
File: go/proto/ctrl/v1/deployment.proto:33-36
Timestamp: 2025-10-15T10:12:40.810Z
Learning: In the Unkey codebase proto files (ctrl/v1/build.proto, ctrl/v1/deployment.proto, hydra/v1/deployment.proto), use `dockerfile_path` (not `docker_file_path`) for consistency in generated Go field names.
Applied to files:
go/Makefile
📚 Learning: 2025-09-12T08:01:20.792Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3944
File: go/pkg/db/acme_challenge_update_verified_with_expiry.sql_generated.go:31-39
Timestamp: 2025-09-12T08:01:20.792Z
Learning: Do not review or suggest changes to files with sql_generated.go suffix or other files marked as auto-generated (containing "Code generated by" comments), as these are generated by tools like sqlc and changes would be overwritten on regeneration.
Applied to files:
go/go.mod
📚 Learning: 2025-07-15T14:59:30.212Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3560
File: go/deploy/metald/internal/database/repository.go:0-0
Timestamp: 2025-07-15T14:59:30.212Z
Learning: go/deploy/metald cannot currently import helpers from go/pkg/db because it is not yet part of the main Go module; avoid suggesting such imports until the modules are unified.
Applied to files:
go/go.mod
📚 Learning: 2025-04-08T09:34:24.576Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 2872
File: apps/dashboard/lib/trpc/routers/ratelimit/createNamespace.ts:36-39
Timestamp: 2025-04-08T09:34:24.576Z
Learning: In the Unkey dashboard, when making database queries involving workspaces, use `ctx.workspace.id` directly instead of fetching the workspace separately for better performance and security.
Applied to files:
go/apps/api/routes/v2_analytics_get_verifications/handler.go
📚 Learning: 2025-04-08T09:34:24.576Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 2872
File: apps/dashboard/lib/trpc/routers/ratelimit/createNamespace.ts:36-39
Timestamp: 2025-04-08T09:34:24.576Z
Learning: When querying or updating namespaces in the Unkey dashboard, always scope the operations to the current workspace using `eq(table.workspaceId, ctx.workspace.id)` to prevent cross-workspace access.
Applied to files:
go/apps/api/routes/v2_analytics_get_verifications/handler.go
📚 Learning: 2025-09-01T08:29:10.199Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3895
File: go/pkg/db/key_list_live_by_auth_id.sql_generated.go:98-105
Timestamp: 2025-09-01T08:29:10.199Z
Learning: In Unkey's ListLiveKeysByKeyAuthID query, adding ka.workspace_id = k.workspace_id constraint negatively impacts index performance. Workspace validation is handled upstream at the API level before the query is called, making the additional constraint unnecessary.
Applied to files:
go/apps/api/routes/v2_analytics_get_verifications/handler.go
📚 Learning: 2025-09-22T18:44:56.279Z
Learnt from: perkinsjr
Repo: unkeyed/unkey PR: 4009
File: apps/dashboard/app/(app)/[workspace]/apis/[apiId]/_overview/components/table/components/log-details/index.tsx:4-5
Timestamp: 2025-09-22T18:44:56.279Z
Learning: In the Unkey dashboard, the workspace hook (useWorkspace) provides security validation by checking database access and user authorization to the workspace, with 10-minute caching for performance. Using URL params (useParams) for workspace slug would bypass this security validation and allow unauthorized access attempts. Always use the workspace hook for workspace-scoped navigation and handle loading states properly rather than switching to URL parameters.
Applied to files:
go/apps/api/routes/v2_analytics_get_verifications/handler.go
📚 Learning: 2025-09-15T20:45:05.696Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3952
File: go/apps/ctrl/services/routing/service.go:69-91
Timestamp: 2025-09-15T20:45:05.696Z
Learning: In Unkey's routing service, gateway lookups should be workspace-scoped using FindGatewayByHostnameAndWorkspace instead of hostname-only queries to prevent cross-tenant access issues.
Applied to files:
go/apps/api/routes/v2_analytics_get_verifications/handler.go
📚 Learning: 2025-10-30T15:10:52.743Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 4190
File: go/internal/services/keys/verifier.go:51-53
Timestamp: 2025-10-30T15:10:52.743Z
Learning: PR #4190 for unkeyed/unkey is focused solely on database schema and query changes for identity-based credits. It adds IdentityCredits and KeyCredits fields to structs and queries, but does not implement the priority enforcement logic in the usagelimiter. The logic implementation is intentionally deferred to a later PR in the stack.
Applied to files:
go/pkg/clickhouse/key_verifications_test.go
📚 Learning: 2025-10-21T09:45:47.560Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 4107
File: go/pkg/clickhouse/key_verifications_test.go:20-20
Timestamp: 2025-10-21T09:45:47.560Z
Learning: In ClickHouse tests (e.g., go/pkg/clickhouse/key_verifications_test.go), parallel execution with t.Parallel() is safe when tests use workspaceID-based isolation. Each test generates a unique workspaceID (via uid.New(uid.WorkspacePrefix)), and all database operations are scoped to that workspaceID, providing logical isolation even when tests share a single ClickHouse instance.
Applied to files:
go/pkg/clickhouse/key_verifications_test.go
📚 Learning: 2025-08-27T13:48:54.016Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3841
File: go/pkg/db/queries/key_find_for_verification.sql:15-15
Timestamp: 2025-08-27T13:48:54.016Z
Learning: The pending_migration_id field in FindKeyForVerification is only used internally by get_migrated.go for migration logic and doesn't need to be exposed in KeyVerifier struct or API response DTOs since it's an internal implementation detail.
Applied to files:
go/pkg/clickhouse/key_verifications_test.go
📚 Learning: 2025-07-03T05:58:10.699Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3421
File: go/apps/api/openapi/openapi.yaml:196-200
Timestamp: 2025-07-03T05:58:10.699Z
Learning: In the Unkey codebase, OpenAPI 3.1 is used, which allows sibling keys (such as `description`) alongside `$ref` in schema objects. Do not flag this as an error in future reviews.
Applied to files:
go/apps/api/openapi/openapi-generated.yaml
📚 Learning: 2025-09-12T18:11:33.481Z
Learnt from: mcstepp
Repo: unkeyed/unkey PR: 3952
File: apps/dashboard/lib/trpc/routers/deploy/rollback.ts:23-24
Timestamp: 2025-09-12T18:11:33.481Z
Learning: In apps/dashboard/lib/trpc/routers/deploy/rollback.ts, the CTRL_URL environment variable should fail fast with a clear error message if missing in non-development environments, rather than defaulting to localhost which can mask production configuration issues.
Applied to files:
go/apps/api/openapi/openapi-generated.yaml
📚 Learning: 2025-02-24T13:35:00.816Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 2910
File: apps/engineering/content/design/components/form/form-input.variants.tsx:38-38
Timestamp: 2025-02-24T13:35:00.816Z
Learning: In the Unkey codebase, example values in documentation and demo components intentionally use themed fake data (e.g., Middle Earth references) rather than generic placeholder text.
Applied to files:
apps/docs/docs.json
🧬 Code graph analysis (13)
go/internal/services/caches/caches.go (7)
go/pkg/cache/interface.go (1)
Cache(7-49)go/pkg/db/models_generated.go (1)
ClickhouseWorkspaceSetting(569-581)go/pkg/cache/scoped_key.go (3)
ScopedKey(49-62)ScopedKeyToString(81-81)ScopedKeyFromString(82-82)go/pkg/db/api_find_key_auth_by_key_auth_ids.sql_generated.go (1)
FindKeyAuthsByKeyAuthIdsRow(28-31)go/pkg/db/api_find_key_auth_by_ids.sql_generated.go (1)
FindKeyAuthsByIdsRow(28-31)go/pkg/cache/cache.go (1)
Config(34-51)go/pkg/cache/middleware/tracing.go (1)
WithTracing(16-18)
go/apps/api/run.go (4)
go/internal/services/analytics/connection_manager.go (3)
NewNoopConnectionManager(192-194)NewConnectionManager(47-85)ConnectionManagerConfig(37-44)go/apps/api/routes/register.go (1)
Register(62-581)go/apps/api/routes/services.go (1)
Services(17-30)go/internal/services/caches/caches.go (1)
Caches(21-49)
go/pkg/zen/middleware_errors.go (2)
go/pkg/codes/constants_gen.go (12)
UnkeyDataErrorsKeySpaceNotFound(93-93)UserErrorsBadRequestInvalidAnalyticsQuery(24-24)UserErrorsBadRequestInvalidAnalyticsTable(26-26)UserErrorsBadRequestInvalidAnalyticsFunction(28-28)UserErrorsBadRequestInvalidAnalyticsQueryType(30-30)UserErrorsUnprocessableEntityQueryExecutionTimeout(35-35)UserErrorsUnprocessableEntityQueryMemoryLimitExceeded(37-37)UserErrorsUnprocessableEntityQueryRowsLimitExceeded(39-39)UserErrorsTooManyRequestsQueryQuotaExceeded(44-44)UnkeyDataErrorsAnalyticsNotConfigured(141-141)UnkeyAppErrorsPreconditionPreconditionFailed(171-171)UnkeyDataErrorsAnalyticsConnectionFailed(143-143)go/apps/api/openapi/gen.go (8)
BadRequestErrorResponse(65-71)Meta(279-282)BadRequestErrorDetails(47-62)ValidationError(2015-2029)UnprocessableEntityErrorResponse(500-506)BaseError(74-86)TooManyRequestsErrorResponse(471-477)InternalServerErrorResponse(181-187)
go/apps/api/openapi/gen.go (1)
go/pkg/codes/unkey_data.go (1)
Data(112-165)
go/pkg/testutil/http.go (5)
apps/agent/services/vault/service.go (2)
Service(20-30)New(39-74)go/internal/services/analytics/connection_manager.go (3)
ConnectionManager(22-24)NewConnectionManager(47-85)ConnectionManagerConfig(37-44)go/gen/proto/vault/v1/service.pb.go (3)
EncryptRequest(104-110)EncryptRequest(123-123)EncryptRequest(138-140)go/pkg/clickhouse/user.go (2)
UserConfig(11-28)DefaultAllowedTables(143-152)go/pkg/db/clickhouse_workspace_settings_insert.sql_generated.go (1)
InsertClickhouseWorkspaceSettingsParams(42-54)
go/pkg/cache/clustering/cluster_cache.go (1)
go/pkg/cache/interface.go (2)
CacheHit(55-55)Op(66-66)
go/pkg/clickhouse/query-parser/cte.go (1)
go/pkg/clickhouse/query-parser/types.go (1)
Parser(23-27)
go/pkg/codes/generate.go (2)
go/pkg/codes/user_request.go (1)
User(53-72)go/pkg/codes/unkey_data.go (1)
Data(112-165)
go/pkg/cache/cache.go (2)
go/pkg/cache/interface.go (8)
CacheHit(55-55)Miss(63-63)Hit(61-61)Null(59-59)Op(66-66)WriteValue(72-72)WriteNull(75-75)Noop(70-70)go/pkg/prometheus/metrics/cache.go (1)
CacheRevalidations(103-112)
go/main.go (3)
go/cmd/create-clickhouse-user/main.go (1)
Cmd(20-65)go/cmd/dev/main.go (1)
Cmd(8-15)go/cmd/dev/seed/run.go (1)
Cmd(7-13)
go/apps/api/routes/v2_analytics_get_verifications/handler.go (16)
go/apps/api/openapi/gen.go (4)
V2AnalyticsGetVerificationsRequestBody(535-539)V2AnalyticsGetVerificationsResponseBody(542-548)V2AnalyticsGetVerificationsResponseData(551-551)Meta(279-282)go/pkg/clickhouse/interface.go (1)
ClickHouse(59-68)go/internal/services/analytics/connection_manager.go (1)
ConnectionManager(22-24)go/internal/services/caches/caches.go (2)
Caches(21-49)Config(62-74)go/pkg/zen/request_util.go (1)
BindBody(10-23)go/pkg/clickhouse/query-parser/types.go (2)
Config(14-20)SecurityFilter(8-11)go/pkg/rbac/permissions.go (5)
Limit(77-77)Tuple(178-187)ResourceType(11-11)Api(21-21)ReadAnalytics(70-70)go/internal/services/keys/options.go (1)
WithPermissions(47-52)go/pkg/clickhouse/errors.go (1)
WrapClickHouseError(219-253)go/pkg/cache/scoped_key.go (1)
ScopedKey(49-62)go/pkg/db/api_find_key_auth_by_ids.sql_generated.go (2)
FindKeyAuthsByIdsRow(28-31)FindKeyAuthsByIdsParams(23-26)go/pkg/array/map.go (1)
Map(23-29)go/pkg/cache/interface.go (3)
Key(51-53)Null(59-59)CacheHit(55-55)go/pkg/array/reduce.go (1)
Reduce(28-34)go/internal/services/caches/op.go (1)
DefaultFindFirstOp(9-22)go/pkg/db/api_find_key_auth_by_key_auth_ids.sql_generated.go (2)
FindKeyAuthsByKeyAuthIdsRow(28-31)FindKeyAuthsByKeyAuthIdsParams(23-26)
go/pkg/cache/middleware/tracing.go (1)
go/pkg/cache/interface.go (3)
CacheHit(55-55)Miss(63-63)Op(66-66)
go/pkg/clickhouse/key_verifications_test.go (4)
go/pkg/array/random.go (1)
Random(23-29)go/pkg/clickhouse/schema/types.go (1)
KeyVerificationV2(6-19)go/pkg/uid/uid.go (1)
RequestPrefix(19-19)go/pkg/array/reduce.go (1)
Reduce(28-34)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (5)
- GitHub Check: Test Dashboard / Test Dashboard
- GitHub Check: Test Go API Local / Test
- GitHub Check: Test API / API Test Local
- GitHub Check: Build / Build
- GitHub Check: Test Packages / Test
🔇 Additional comments (26)
go/Makefile (1)
27-27: Good:--waitflag ensures services are ready before proceeding.Adding
--waitto the docker compose command is a helpful improvement for the development workflow, ensuring all services are fully available before downstream steps execute.apps/docs/docs.json (3)
171-181: Analytics integration looks solid.The new analytics pages (overview, getting-started, query-examples, schema-reference, query-restrictions) are properly integrated into the Features navigation, and the analytics error pages are correctly categorized under Data errors and the new User Errors groupings. Aligns well with the PR's new
/v2/analytics.getVerificationsendpoint.
257-281: User Errors reorganization is clean.Breaking user-facing errors into Bad Request, Too Many Requests, and Unprocessable Entity groups is sensible and the analytics-specific errors (invalid_analytics_query, invalid_analytics_table, invalid_analytics_function) map correctly to 400-level codes. Query quota/timeout/memory/rows limit errors properly categorized.
172-172: Clarify the explicit"hidden": falseon Analytics.This is the only navigation group with an explicit
"hidden": falsefield (line 172). If this is intentional to keep it visible by default, that's fine, but if it's residual from an earlier iteration, it can be removed sincefalseis the Mintlify default.go/go.mod (3)
42-42: Clarify the sqlc version downgrade.The sqlc dependency was downgraded from
v1.30.0tov1.29.0. This is unusual and the PR description does not explain the reason. Confirm that:
- This downgrade is intentional and not accidental.
- There is no compatibility issue with v1.30.0.
- Any code generation artifacts (e.g.,
*_generated.gofiles) remain compatible.Please provide context for why sqlc was downgraded. If this is a workaround for a bug in v1.30.0, document it inline as you've done for other replace blocks.
334-338: Verify necessity of replace blocks for go-yit and pingcap/tidb/pkg/parser.Two replace blocks have been added to work around YAML parsing and SQLC engine errors. These are workarounds for upstream issues. Confirm:
- These pinned versions are the minimal necessary to resolve the issues.
- There is a plan to remove these replaces when upstream issues are resolved.
- The specific commit hashes or pseudo-versions chosen are correct and don't introduce their own issues.
Consider adding tracking comments or links to upstream issues in the replace directives to aid future maintenance.
11-11: Document the SQLC downgrade reason; ClickHouse SQL parser is safe to add.SQLC v1.30.0 is mainly a bug fix release with a security fix for output plugins, but includes BREAKING changes for MySQL type mappings (int16 for SMALLINT/YEAR, int8 for TINYINT). Add a comment in
go.modexplaining why the downgrade from v1.30.0 to v1.29.0 was necessary. The new ClickHouse SQL parser dependency is safe—no known vulnerabilities and compatible with the existing ClickHouse Go driver v2.40.1.go/pkg/clickhouse/query-parser/cte.go (2)
7-26: Excellent documentation addressing previous feedback.The comprehensive comment effectively explains what CTEs are, why tracking is necessary, and provides a clear example. This directly addresses the past review feedback requesting clarification.
27-41: The concern about nil map panic is unfounded.The
cteNamesfield is properly initialized in theNewParserconstructor withmake(map[string]bool)at parser.go line 16. The code is safe—no nil panic risk exists.Likely an incorrect or invalid review comment.
go/pkg/codes/generate.go (4)
53-84: LGTM! Clean end-to-end MDX generation workflow.The main function correctly collects error codes from all domains and orchestrates the MDX generation, cleanup, and docs.json update with appropriate error handling.
86-92: LGTM! Well-defined metadata structure.The ErrorCodeInfo struct appropriately captures all necessary information for MDX and docs.json generation.
157-182: LGTM! Clean signature update.The processErrorDomain function correctly aggregates error codes from all categories and returns them for MDX generation.
186-236: LGTM! Proper metadata extraction and collection.The processCategory function correctly extracts descriptions from comments and builds ErrorCodeInfo entries for each error code.
go/pkg/cache/interface.go (1)
12-36: LGTM! Clean interface extension for bulk operations.The multi-key methods follow consistent naming conventions and signatures. The documentation is clear, and the method signatures properly handle batch operations by using slices for input keys and maps for returning results indexed by key.
go/pkg/zen/middleware_errors.go (6)
40-40: LGTM! Correct NotFound error mapping.Adding
KeySpaceNotFoundto the NotFound error block is appropriate and follows the existing pattern for resource-not-found errors.
61-61: Good clarification.The updated comment distinguishes general validation errors from the new analytics-specific Bad Request errors, improving code organization.
81-97: LGTM! Correct Bad Request mapping for analytics queries.The analytics query validation errors correctly return 400 status with
BadRequestErrorResponse. The error codes align with the constants defined ingo/pkg/codes/constants_gen.go.
99-113: LGTM! Appropriate Unprocessable Entity mapping.Query resource limit errors (timeout, memory, rows) correctly return 422 status, distinguishing them from malformed query errors (400). The response type matches the OpenAPI schema.
115-127: LGTM! Correct rate limiting error mapping.The quota exceeded error appropriately returns 429 status with
TooManyRequestsErrorResponse. Note: PR objectives mention removing max-rows-to-read quota as a follow-up item.
262-263: LGTM! Appropriate precondition mapping.Adding
AnalyticsNotConfiguredto the Precondition Failed block is correct—analytics must be configured before use.deployment/docker-compose.yaml (4)
74-75: LGTM! Correct dependency addition.Adding s3 as a dependency with health check for the apiv2 service is appropriate, as the analytics feature requires vault access for storing encrypted ClickHouse credentials.
89-93: LGTM! Consistent environment variable naming.The vault-related environment variables now follow the
UNKEY_VAULT_*naming convention, improving consistency across the configuration.
95-95: LGTM! Correct analytics URL configuration.The
UNKEY_CLICKHOUSE_ANALYTICS_URLcorrectly points to the ClickHouse HTTP interface using the Docker network service name and standard port.
126-131: LGTM! Appropriate healthcheck configuration.The Kafka healthcheck using
kafka-topics.sh --listis a good way to verify the broker is operational, and the timing parameters are reasonable.go/apps/api/openapi/openapi-generated.yaml (1)
170-187: 422 description may conflict with “remove max-rows-to-read quota” follow-upYou mention “scans too many rows” here. Chronark’s follow-up asked to remove max-rows-to-read quota. Ensure the implementation and docs match; if that quota was removed, adjust this description accordingly.
go/main.go (1)
36-38: CLI wiring LGTM; confirm dev command exposure in releasesclickhouseUser.Cmd and dev.Cmd are correctly registered. Please confirm the dev tooling should ship in production binaries; if not, gate via build tags or an env flag.
There was a problem hiding this comment.
Actionable comments posted: 3
🧹 Nitpick comments (1)
go/pkg/testutil/http.go (1)
256-290: Consider adding option builders for all configurable fieldsThe options pattern is well-implemented, but option builders are only provided for 4 out of 6 configurable fields. Consider adding:
WithQuotaDurationSecondsWithMaxQueryExecutionTimeThis would provide test authors full control over analytics quota configuration when needed for edge-case testing.
Add the missing option builders:
+func WithQuotaDurationSeconds(seconds int32) SetupAnalyticsOption { + return func(c *setupAnalyticsConfig) { + c.QuotaDurationSeconds = seconds + } +} + +func WithMaxQueryExecutionTime(seconds int32) SetupAnalyticsOption { + return func(c *setupAnalyticsConfig) { + c.MaxQueryExecutionTime = seconds + } +}
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (8)
apps/docs/analytics/query-examples.mdx(1 hunks)go/apps/api/routes/v2_analytics_get_verifications/422_test.go(1 hunks)go/internal/services/caches/caches.go(2 hunks)go/pkg/clickhouse/key_verifications_test.go(4 hunks)go/pkg/clickhouse/query-parser/cte.go(1 hunks)go/pkg/clickhouse/user.go(1 hunks)go/pkg/testutil/http.go(4 hunks)go/pkg/zen/middleware_errors.go(5 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- go/pkg/clickhouse/query-parser/cte.go
🧰 Additional context used
🧠 Learnings (13)
📓 Common learnings
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 4190
File: go/internal/services/keys/verifier.go:51-53
Timestamp: 2025-10-30T15:10:52.743Z
Learning: PR #4190 for unkeyed/unkey is focused solely on database schema and query changes for identity-based credits. It adds IdentityCredits and KeyCredits fields to structs and queries, but does not implement the priority enforcement logic in the usagelimiter. The logic implementation is intentionally deferred to a later PR in the stack.
📚 Learning: 2025-04-08T09:34:24.576Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 2872
File: apps/dashboard/lib/trpc/routers/ratelimit/createNamespace.ts:36-39
Timestamp: 2025-04-08T09:34:24.576Z
Learning: In the Unkey dashboard, when making database queries involving workspaces, use `ctx.workspace.id` directly instead of fetching the workspace separately for better performance and security.
Applied to files:
go/pkg/clickhouse/user.go
📚 Learning: 2025-10-21T09:45:47.560Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 4107
File: go/pkg/clickhouse/key_verifications_test.go:20-20
Timestamp: 2025-10-21T09:45:47.560Z
Learning: In ClickHouse tests (e.g., go/pkg/clickhouse/key_verifications_test.go), parallel execution with t.Parallel() is safe when tests use workspaceID-based isolation. Each test generates a unique workspaceID (via uid.New(uid.WorkspacePrefix)), and all database operations are scoped to that workspaceID, providing logical isolation even when tests share a single ClickHouse instance.
Applied to files:
go/pkg/clickhouse/user.gogo/pkg/clickhouse/key_verifications_test.go
📚 Learning: 2025-04-08T09:34:24.576Z
Learnt from: ogzhanolguncu
Repo: unkeyed/unkey PR: 2872
File: apps/dashboard/lib/trpc/routers/ratelimit/createNamespace.ts:36-39
Timestamp: 2025-04-08T09:34:24.576Z
Learning: When querying or updating namespaces in the Unkey dashboard, always scope the operations to the current workspace using `eq(table.workspaceId, ctx.workspace.id)` to prevent cross-workspace access.
Applied to files:
go/pkg/clickhouse/user.go
📚 Learning: 2025-09-22T18:44:56.279Z
Learnt from: perkinsjr
Repo: unkeyed/unkey PR: 4009
File: apps/dashboard/app/(app)/[workspace]/apis/[apiId]/_overview/components/table/components/log-details/index.tsx:4-5
Timestamp: 2025-09-22T18:44:56.279Z
Learning: In the Unkey dashboard, the workspace hook (useWorkspace) provides security validation by checking database access and user authorization to the workspace, with 10-minute caching for performance. Using URL params (useParams) for workspace slug would bypass this security validation and allow unauthorized access attempts. Always use the workspace hook for workspace-scoped navigation and handle loading states properly rather than switching to URL parameters.
Applied to files:
go/pkg/clickhouse/user.go
📚 Learning: 2025-10-30T15:10:52.743Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 4190
File: go/internal/services/keys/verifier.go:51-53
Timestamp: 2025-10-30T15:10:52.743Z
Learning: PR #4190 for unkeyed/unkey is focused solely on database schema and query changes for identity-based credits. It adds IdentityCredits and KeyCredits fields to structs and queries, but does not implement the priority enforcement logic in the usagelimiter. The logic implementation is intentionally deferred to a later PR in the stack.
Applied to files:
go/pkg/clickhouse/user.gogo/pkg/clickhouse/key_verifications_test.go
📚 Learning: 2025-08-14T16:25:48.167Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3785
File: go/apps/api/routes/v2_keys_reroll_key/401_test.go:52-61
Timestamp: 2025-08-14T16:25:48.167Z
Learning: User Flo4604 requested creation of a GitHub issue to track converting all test files to use table-driven test patterns as a broader codebase improvement, following the suggestion made during review of go/apps/api/routes/v2_keys_reroll_key/401_test.go.
Applied to files:
go/pkg/clickhouse/user.gogo/pkg/clickhouse/key_verifications_test.gogo/apps/api/routes/v2_analytics_get_verifications/422_test.go
📚 Learning: 2025-07-15T14:25:05.608Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3560
File: go/apps/api/routes/v2_keys_create_key/handler.go:353-466
Timestamp: 2025-07-15T14:25:05.608Z
Learning: In the Unkey codebase, input validation for API endpoints is handled at the OpenAPI schema layer, which validates request fields like permission slugs (pattern: "^[a-zA-Z0-9_]+$", length: 1-100 characters) before requests reach the handler code. This validation occurs during the zen.BindBody call in handlers.
Applied to files:
go/pkg/clickhouse/user.gogo/pkg/zen/middleware_errors.go
📚 Learning: 2025-08-27T13:48:54.016Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3841
File: go/pkg/db/queries/key_find_for_verification.sql:15-15
Timestamp: 2025-08-27T13:48:54.016Z
Learning: The pending_migration_id field in FindKeyForVerification is only used internally by get_migrated.go for migration logic and doesn't need to be exposed in KeyVerifier struct or API response DTOs since it's an internal implementation detail.
Applied to files:
go/pkg/clickhouse/key_verifications_test.go
📚 Learning: 2025-07-16T15:38:53.491Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3606
File: go/pkg/db/replica.go:8-11
Timestamp: 2025-07-16T15:38:53.491Z
Learning: For debugging database replica usage in go/pkg/db/replica.go, it's acceptable to mark QueryRowContext operations as "success" even though SQL errors only surface during row.Scan() calls. The timing metrics are the primary concern for debugging replica performance patterns.
Applied to files:
go/pkg/clickhouse/key_verifications_test.gogo/pkg/testutil/http.go
📚 Learning: 2025-09-15T19:53:28.487Z
Learnt from: mcstepp
Repo: unkeyed/unkey PR: 3952
File: go/proto/ctrl/v1/routing.proto:0-0
Timestamp: 2025-09-15T19:53:28.487Z
Learning: In the Unkey codebase, authentication/authorization errors intentionally return NOT_FOUND error codes instead of distinct auth error codes (like FORBIDDEN or UNAUTHORIZED) for security reasons. This prevents attackers from distinguishing between "resource doesn't exist" and "you don't have permission to access this resource", avoiding information disclosure about workspace existence and access boundaries.
Applied to files:
go/pkg/zen/middleware_errors.go
📚 Learning: 2025-03-19T09:25:59.751Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 2955
File: go/apps/api/routes/v2_identities_create_identity/handler.go:162-202
Timestamp: 2025-03-19T09:25:59.751Z
Learning: In the Unkey codebase, input validation for API endpoints is primarily handled through OpenAPI schema validation, which occurs before requests reach the handler code. For example, in the identities.createIdentity endpoint, minimum values for ratelimit duration and limit are defined in the OpenAPI schema rather than duplicating these checks in the handler.
Applied to files:
go/pkg/zen/middleware_errors.go
📚 Learning: 2025-09-12T13:25:41.849Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3958
File: internal/clickhouse/src/requests.ts:7-8
Timestamp: 2025-09-12T13:25:41.849Z
Learning: In the ClickHouse v2 table migration, the location field was renamed from `colo` to `region` across all tables including `default.api_requests_raw_v2`. The zod insert schemas must be updated accordingly to avoid runtime insert failures.
Applied to files:
apps/docs/analytics/query-examples.mdx
🧬 Code graph analysis (5)
go/internal/services/caches/caches.go (7)
go/pkg/db/models_generated.go (1)
ClickhouseWorkspaceSetting(569-581)go/pkg/cache/scoped_key.go (3)
ScopedKey(49-62)ScopedKeyToString(81-81)ScopedKeyFromString(82-82)go/pkg/db/api_find_key_auth_by_key_auth_ids.sql_generated.go (1)
FindKeyAuthsByKeyAuthIdsRow(28-31)go/pkg/db/api_find_key_auth_by_ids.sql_generated.go (1)
FindKeyAuthsByIdsRow(28-31)go/pkg/cache/clustering/cluster_cache.go (1)
Config(34-58)go/pkg/cache/cache.go (1)
Config(34-51)go/pkg/cache/middleware/tracing.go (1)
WithTracing(16-18)
go/pkg/clickhouse/key_verifications_test.go (4)
go/pkg/clickhouse/client.go (1)
New(62-208)go/pkg/array/random.go (1)
Random(23-29)go/pkg/clickhouse/schema/types.go (1)
KeyVerificationV2(6-19)go/pkg/array/reduce.go (1)
Reduce(28-34)
go/pkg/testutil/http.go (6)
go/internal/services/caches/caches.go (2)
Caches(21-49)New(166-313)go/internal/services/analytics/connection_manager.go (3)
ConnectionManager(22-24)NewConnectionManager(47-85)ConnectionManagerConfig(37-44)go/gen/proto/vault/v1/service.pb.go (3)
EncryptRequest(104-110)EncryptRequest(123-123)EncryptRequest(138-140)go/pkg/clickhouse/user.go (1)
UserConfig(22-39)go/pkg/db/clickhouse_workspace_settings_insert.sql_generated.go (1)
InsertClickhouseWorkspaceSettingsParams(42-54)go/gen/proto/vault/v1/object.pb.go (3)
Encrypted(252-264)Encrypted(277-277)Encrypted(292-294)
go/pkg/zen/middleware_errors.go (2)
go/pkg/codes/constants_gen.go (12)
UnkeyDataErrorsKeySpaceNotFound(93-93)UserErrorsBadRequestInvalidAnalyticsQuery(24-24)UserErrorsBadRequestInvalidAnalyticsTable(26-26)UserErrorsBadRequestInvalidAnalyticsFunction(28-28)UserErrorsBadRequestInvalidAnalyticsQueryType(30-30)UserErrorsUnprocessableEntityQueryExecutionTimeout(35-35)UserErrorsUnprocessableEntityQueryMemoryLimitExceeded(37-37)UserErrorsUnprocessableEntityQueryRowsLimitExceeded(39-39)UserErrorsTooManyRequestsQueryQuotaExceeded(44-44)UnkeyDataErrorsAnalyticsNotConfigured(141-141)UnkeyAppErrorsPreconditionPreconditionFailed(171-171)UnkeyDataErrorsAnalyticsConnectionFailed(143-143)go/apps/api/openapi/gen.go (8)
BadRequestErrorResponse(65-71)Meta(279-282)BadRequestErrorDetails(47-62)ValidationError(2015-2029)UnprocessableEntityErrorResponse(500-506)BaseError(74-86)TooManyRequestsErrorResponse(471-477)InternalServerErrorResponse(181-187)
go/apps/api/routes/v2_analytics_get_verifications/422_test.go (6)
go/pkg/testutil/http.go (3)
NewHarness(59-208)WithMaxQueryMemoryBytes(273-277)CallRoute(385-419)go/pkg/testutil/seed/seed.go (1)
CreateApiRequest(84-92)go/pkg/clickhouse/interface.go (1)
ClickHouse(59-68)go/pkg/clickhouse/schema/requests.go (1)
KeyVerificationRequestV1(66-92)go/pkg/uid/uid.go (2)
RequestPrefix(19-19)KeyPrefix(16-16)go/apps/api/routes/v2_analytics_get_verifications/handler.go (3)
Handler(49-56)Request(26-26)Response(27-27)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (6)
- GitHub Check: Test Packages / Test
- GitHub Check: Test Go API Local / Test
- GitHub Check: Test Dashboard / Test Dashboard
- GitHub Check: Test API / API Test Local
- GitHub Check: Build / Build
- GitHub Check: Analyze (javascript-typescript)
🔇 Additional comments (20)
apps/docs/analytics/query-examples.mdx (2)
628-657: ✅ Moving average example table reference fixed.The previous critical inconsistency between readable SQL (line 642) and JSON format (line 653) has been resolved. Both now correctly reference
key_verifications_per_day_v1, ensuring users copying either format will get working queries.
1-732: Well‑structured analytics documentation with consistent examples.The file provides comprehensive, well‑organized ClickHouse SQL examples across all key analytics scenarios. Each example clearly shows both readable multi-line and single-line JSON formats for API requests. All ~28 query examples maintain consistency between formats, use correct table names with _v1 versioning, and include appropriate time filters.
The documentation aligns well with the new
/v2/analytics.getVerificationsendpoint and should effectively guide users in writing analytics queries.go/apps/api/routes/v2_analytics_get_verifications/422_test.go (1)
30-42: Fix the invalid range loop.Line 30 uses
for i := range 50_000, which is illegal because Go’srangeonly works on strings, arrays, slices, maps, or channels—this code will not compile. Please switch to a standard counted loop.- for i := range 50_000 { + for i := 0; i < 50_000; i++ {⛔ Skipped due to learnings
Learnt from: Flo4604 Repo: unkeyed/unkey PR: 3785 File: go/apps/api/routes/v2_keys_reroll_key/401_test.go:52-61 Timestamp: 2025-08-14T16:25:48.167Z Learning: User Flo4604 requested creation of a GitHub issue to track converting all test files to use table-driven test patterns as a broader codebase improvement, following the suggestion made during review of go/apps/api/routes/v2_keys_reroll_key/401_test.go.Learnt from: Flo4604 Repo: unkeyed/unkey PR: 3800 File: go/apps/api/integration/multi_node_usagelimiting/run.go:107-126 Timestamp: 2025-08-19T08:57:31.793Z Learning: Go 1.22+ supports "range over integers" syntax: `for range N` iterates N times, and `for i := range N` iterates with i from 0 to N-1. This is valid Go syntax and should not be flagged as a compilation error.Learnt from: imeyer Repo: unkeyed/unkey PR: 3733 File: go/pkg/repeat/every_test.go:383-383 Timestamp: 2025-08-05T15:19:50.563Z Learning: Go 1.24 introduced testing.B.Loop method which simplifies benchmark functions by handling iteration internally, ensuring setup/cleanup runs only once per benchmark count, and preventing compiler optimizations that could affect benchmark accuracy.Learnt from: ogzhanolguncu Repo: unkeyed/unkey PR: 4107 File: go/pkg/clickhouse/key_verifications_test.go:20-20 Timestamp: 2025-10-21T09:45:47.560Z Learning: In ClickHouse tests (e.g., go/pkg/clickhouse/key_verifications_test.go), parallel execution with t.Parallel() is safe when tests use workspaceID-based isolation. Each test generates a unique workspaceID (via uid.New(uid.WorkspacePrefix)), and all database operations are scoped to that workspaceID, providing logical isolation even when tests share a single ClickHouse instance.go/pkg/zen/middleware_errors.go (4)
40-40: LGTM - KeySpaceNotFound error mappingThe addition of
UnkeyDataErrorsKeySpaceNotFoundto the Not Found error group follows the established pattern and is appropriate for the analytics feature.
61-97: LGTM - Analytics query validation error handlingThe split into "General validation" and "Query validation" categories improves code organization. All analytics query validation errors correctly return 400 Bad Request with consistent error response structures.
99-113: LGTM - Query resource limit error handlingUsing 422 Unprocessable Entity for query resource limits (timeout, memory, rows) is semantically appropriate and follows HTTP best practices.
115-127: LGTM - Query quota exceeded error handlingUsing 429 Too Many Requests for quota exceeded is the correct HTTP status code for rate limiting scenarios.
go/internal/services/caches/caches.go (5)
34-44: LGTM - New cache field declarationsThe three new cache fields are well-documented with clear descriptions of their key-value types and purposes. The choice of key types (string for workspace-scoped settings, ScopedKey for workspace+resource pairs) is appropriate for each use case.
248-264: LGTM - ClickhouseSetting cache initializationThe cache configuration is appropriate with 1-minute freshness for workspace settings that don't change frequently. Error handling is correctly implemented.
266-283: LGTM - KeyAuthToApiRow cache initializationThe 10-minute freshness window is appropriate for key_auth to API mappings that are relatively stable. Proper use of ScopedKey converters and error handling.
285-302: LGTM - ApiToKeyAuthRow cache initialization with error handlingThe cache initialization follows the established pattern with proper error handling at lines 300-302. The configuration mirrors KeyAuthToApiRow appropriately.
308-310: LGTM - Cache wiring with tracingThe new caches are correctly wrapped with tracing middleware and added to the returned Caches struct, maintaining consistency with existing cache initialization.
go/pkg/clickhouse/user.go (5)
11-19: LGTM - Secure identifier validation patternsThe regex patterns properly restrict identifiers to alphanumeric characters and underscores, with proper anchoring (^ and $) to prevent SQL injection attacks.
21-39: LGTM - UserConfig structureThe struct is well-organized with clear separation between per-window quotas and per-query limits. All necessary configuration fields are present for comprehensive user provisioning.
41-62: LGTM - Comprehensive identifier validationThis validation function effectively addresses SQL injection concerns by ensuring all identifiers (username, workspace ID, table names) match safe patterns before being used in SQL statements. The descriptive error messages aid in debugging invalid configurations.
64-171: ConfigureUser implementation is secure with proper validationThe method implements comprehensive user provisioning with proper security controls:
- SQL injection prevention: All identifiers are validated via
validateIdentifiersbefore use- Password security: Password is bound as a parameter (line 77), not interpolated
- Idempotent operations: Uses
IF NOT EXISTSandOR REPLACEfor safe re-runs- Proper error handling: Errors are wrapped with context and returned appropriately
- Operational logging: Progress is logged at appropriate levels
The workspace_id in the ROW POLICY USING clause (line 106) is string-interpolated but is validated first. This appears to be a ClickHouse limitation—ROW POLICY USING clauses require compile-time constant expressions and don't support parameterization like WHERE clauses do.
Note: Line 124's comment documenting why
max_result_rowsis intentionally omitted from the quota is excellent—it explains that per-window limits are too restrictive for analytics while per-query limits still apply via the settings profile.
173-183: LGTM - Default allowed tablesThe default table list covers the necessary analytics tables with appropriate granularity levels (raw, minute, hour, day, month aggregations).
go/pkg/testutil/http.go (3)
14-15: LGTM - Analytics testing infrastructure additionsThe new imports and AnalyticsConnectionManager field properly extend the test harness to support analytics testing scenarios.
Also applies to: 55-55
164-173: LGTM - Analytics connection manager initializationThe connection manager is properly initialized with the correct ClickHouse DSN from
containers.ClickHouse(t)(line 170), which resolves the previous concern about hard-coded localhost addresses. All required dependencies are correctly wired.
291-350: LGTM - SetupAnalytics implementationThe method correctly orchestrates analytics setup for testing:
- Sensible defaults: 10M rows, 1GB memory, reasonable query/time limits
- Password security: Test password is encrypted via Vault before storage
- Complete provisioning: Configures both ClickHouse (user, permissions, quotas) and MySQL (credential storage)
- Proper error handling: Uses require.NoError to fail fast on setup issues
The hard-coded test password is appropriate for test environments.
There was a problem hiding this comment.
Actionable comments posted: 0
♻️ Duplicate comments (1)
go/apps/api/openapi/openapi-generated.yaml (1)
2047-2058: BadRequest details still unsatisfiable due to BaseError additionalProperties: false
BadRequestErrorDetailsuses allOf to extendBaseErrorwitherrors, butBaseErrordisallows additional properties, soerrorscan never be present. Fix one of:
- Relax
BaseErrorby allowing additional properties, or- Stop using allOf and define a concrete schema that includes the base fields plus
errors.Minimal change:
BaseError: @@ - additionalProperties: false + additionalProperties: trueAlso applies to: 2060-2085
🧹 Nitpick comments (4)
go/apps/api/openapi/openapi-generated.yaml (4)
3348-3423: Document required permissions and confirm visibility
- Add a “Required Permissions” section to match other endpoints and the new RBAC:
api.*.read_analyticsorapi.<api_id>.read_analytics.- Confirm
x-hidden: trueis intentional; otherwise this won’t appear in published SDK/docs.Example addition in description:
description: | Execute custom SQL queries against your key verification analytics. For complete documentation including available tables, columns, data types, query examples, see the schema reference in the API documentation. + + Required Permissions + - `api.*.read_analytics` + - `api.<api_id>.read_analytics`
97-108: Consider basic validation on query sizeAdd
minLength/maxLengthtoquery(e.g.,minLength: 1,maxLength: 20000) to fail fast on pathological payloads and align with backend limits.
2113-2126: Clarify timestamp units in example rowsIf
timeis epoch milliseconds (as example suggests), state it explicitly to avoid SDK/doc ambiguity.
187-205: 429 docs: mention Retry-After headerAdd guidance that responses may include
Retry-Afterand clients should honor it. Helps interoperable backoff behavior.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (4)
go/apps/api/openapi/gen.go(4 hunks)go/apps/api/openapi/openapi-generated.yaml(6 hunks)go/apps/api/openapi/spec/paths/v2/analytics/getVerifications/index.yaml(1 hunks)go/pkg/zen/middleware_errors.go(5 hunks)
🧰 Additional context used
🧠 Learnings (10)
📓 Common learnings
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 4190
File: go/internal/services/keys/verifier.go:51-53
Timestamp: 2025-10-30T15:10:52.743Z
Learning: PR #4190 for unkeyed/unkey is focused solely on database schema and query changes for identity-based credits. It adds IdentityCredits and KeyCredits fields to structs and queries, but does not implement the priority enforcement logic in the usagelimiter. The logic implementation is intentionally deferred to a later PR in the stack.
📚 Learning: 2025-09-15T19:53:28.487Z
Learnt from: mcstepp
Repo: unkeyed/unkey PR: 3952
File: go/proto/ctrl/v1/routing.proto:0-0
Timestamp: 2025-09-15T19:53:28.487Z
Learning: In the Unkey codebase, authentication/authorization errors intentionally return NOT_FOUND error codes instead of distinct auth error codes (like FORBIDDEN or UNAUTHORIZED) for security reasons. This prevents attackers from distinguishing between "resource doesn't exist" and "you don't have permission to access this resource", avoiding information disclosure about workspace existence and access boundaries.
Applied to files:
go/pkg/zen/middleware_errors.gogo/apps/api/openapi/openapi-generated.yaml
📚 Learning: 2025-07-15T14:25:05.608Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3560
File: go/apps/api/routes/v2_keys_create_key/handler.go:353-466
Timestamp: 2025-07-15T14:25:05.608Z
Learning: In the Unkey codebase, input validation for API endpoints is handled at the OpenAPI schema layer, which validates request fields like permission slugs (pattern: "^[a-zA-Z0-9_]+$", length: 1-100 characters) before requests reach the handler code. This validation occurs during the zen.BindBody call in handlers.
Applied to files:
go/pkg/zen/middleware_errors.go
📚 Learning: 2025-03-19T09:25:59.751Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 2955
File: go/apps/api/routes/v2_identities_create_identity/handler.go:162-202
Timestamp: 2025-03-19T09:25:59.751Z
Learning: In the Unkey codebase, input validation for API endpoints is primarily handled through OpenAPI schema validation, which occurs before requests reach the handler code. For example, in the identities.createIdentity endpoint, minimum values for ratelimit duration and limit are defined in the OpenAPI schema rather than duplicating these checks in the handler.
Applied to files:
go/pkg/zen/middleware_errors.go
📚 Learning: 2025-08-08T15:09:01.312Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3753
File: go/apps/api/openapi/config.yaml:9-10
Timestamp: 2025-08-08T15:09:01.312Z
Learning: Repo unkeyed/unkey: For go/apps/api/openapi, oapi-codegen is used and does not support OpenAPI 3.1 union types like [T, "null"]; an overlay step is required to downconvert to 3.0-style nullable before code generation.
Applied to files:
go/apps/api/openapi/gen.go
📚 Learning: 2025-08-08T15:09:01.312Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3753
File: go/apps/api/openapi/config.yaml:9-10
Timestamp: 2025-08-08T15:09:01.312Z
Learning: Repo unkeyed/unkey: In go/apps/api/openapi, oapi-codegen doesn’t support OpenAPI 3.1 union nullability; overlay.yaml must be applied before codegen. The overlay key in oapi-codegen config isn’t supported—use a pre-step (programmatic or CLI) to merge overlay into the bundled spec, then run oapi-codegen.
Applied to files:
go/apps/api/openapi/gen.go
📚 Learning: 2025-08-08T15:09:01.312Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3753
File: go/apps/api/openapi/config.yaml:9-10
Timestamp: 2025-08-08T15:09:01.312Z
Learning: Repo unkeyed/unkey: In go/apps/api/openapi, overlays are required to downconvert 3.1 union nullability to 3.0-style nullable before running oapi-codegen; config.yaml’s output-options.overlay is not recognized by oapi-codegen, so overlays must be applied in a pre-step (programmatic or CLI) prior to codegen.
Applied to files:
go/apps/api/openapi/gen.go
📚 Learning: 2025-08-21T15:54:45.198Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3825
File: go/internal/services/usagelimiter/limit.go:38-0
Timestamp: 2025-08-21T15:54:45.198Z
Learning: In go/internal/services/usagelimiter/limit.go, the UpdateKeyCreditsDecrement operation cannot be safely wrapped with db.WithRetry due to the lack of idempotency mechanisms in the current tech stack. Retrying this non-idempotent write operation risks double-charging users if the first attempt commits but the client sees a transient error.
Applied to files:
go/apps/api/openapi/gen.go
📚 Learning: 2025-07-03T05:58:10.699Z
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 3421
File: go/apps/api/openapi/openapi.yaml:196-200
Timestamp: 2025-07-03T05:58:10.699Z
Learning: In the Unkey codebase, OpenAPI 3.1 is used, which allows sibling keys (such as `description`) alongside `$ref` in schema objects. Do not flag this as an error in future reviews.
Applied to files:
go/apps/api/openapi/openapi-generated.yaml
📚 Learning: 2025-09-12T18:11:33.481Z
Learnt from: mcstepp
Repo: unkeyed/unkey PR: 3952
File: apps/dashboard/lib/trpc/routers/deploy/rollback.ts:23-24
Timestamp: 2025-09-12T18:11:33.481Z
Learning: In apps/dashboard/lib/trpc/routers/deploy/rollback.ts, the CTRL_URL environment variable should fail fast with a clear error message if missing in non-development environments, rather than defaulting to localhost which can mask production configuration issues.
Applied to files:
go/apps/api/openapi/openapi-generated.yaml
🧬 Code graph analysis (1)
go/pkg/zen/middleware_errors.go (2)
go/pkg/codes/constants_gen.go (12)
UnkeyDataErrorsKeySpaceNotFound(93-93)UserErrorsBadRequestInvalidAnalyticsQuery(24-24)UserErrorsBadRequestInvalidAnalyticsTable(26-26)UserErrorsBadRequestInvalidAnalyticsFunction(28-28)UserErrorsBadRequestInvalidAnalyticsQueryType(30-30)UserErrorsUnprocessableEntityQueryExecutionTimeout(35-35)UserErrorsUnprocessableEntityQueryMemoryLimitExceeded(37-37)UserErrorsUnprocessableEntityQueryRowsLimitExceeded(39-39)UserErrorsTooManyRequestsQueryQuotaExceeded(44-44)UnkeyDataErrorsAnalyticsNotConfigured(141-141)UnkeyAppErrorsPreconditionPreconditionFailed(171-171)UnkeyDataErrorsAnalyticsConnectionFailed(143-143)go/apps/api/openapi/gen.go (8)
BadRequestErrorResponse(65-71)Meta(279-282)BadRequestErrorDetails(47-62)ValidationError(2030-2044)UnprocessableEntityErrorResponse(515-521)BaseError(74-86)TooManyRequestsErrorResponse(486-492)ServiceUnavailableErrorResponse(471-477)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (5)
- GitHub Check: Test API / API Test Local
- GitHub Check: Test Packages / Test
- GitHub Check: Build / Build
- GitHub Check: Test Go API Local / Test
- GitHub Check: Test Dashboard / Test Dashboard
🔇 Additional comments (15)
go/apps/api/openapi/spec/paths/v2/analytics/getVerifications/index.yaml (4)
1-11: LGTM!The endpoint configuration is properly structured with appropriate security (rootKey), visibility controls (hidden from docs and SDK generation), and operation naming.
12-15: LGTM!The summary and description are clear and appropriately guide users to consult the schema reference documentation for detailed information about available tables, columns, and query examples.
16-28: LGTM!The request body and success response are properly configured with appropriate schema references and clear descriptions.
29-76: LGTM!The error responses comprehensively cover all expected error scenarios for analytics queries, including query validation (400), resource limits (422), quota enforcement (429), and connectivity issues (503). Each error response has clear descriptions that help users understand and troubleshoot issues.
go/pkg/zen/middleware_errors.go (6)
40-40: LGTM!Adding
KeySpaceNotFoundto the Not Found error group is appropriate and follows the existing pattern for data-related not found errors.
61-97: LGTM!The comment update clarifies that the first Bad Request block handles general validation, while the new analytics query validation block (lines 81-97) appropriately handles query-specific errors. The new error codes comprehensively cover query validation scenarios (invalid queries, tables, functions, and query types) and return the correct 400 Bad Request response.
99-113: LGTM!The Unprocessable Entity (422) error handler appropriately covers query resource limit violations (timeout, memory, rows). The semantic choice of 422 is correct here since the query syntax is valid but the operation cannot be completed due to resource constraints.
115-127: LGTM!The Too Many Requests (429) error handler correctly handles query quota violations, using the appropriate HTTP status code and response type for rate limiting scenarios.
262-263: LGTM!Adding
AnalyticsNotConfiguredto the Precondition Failed (412) error group is semantically correct, as it indicates the analytics feature must be configured before it can be used.
304-323: LGTM!The Service Unavailable (503) error handler correctly addresses analytics connection failures. The code now properly uses
ServiceUnavailableErrorResponse(line 313), which addresses the concern raised in the previous review. The additional logging for this error type is appropriate given the operational significance of service availability issues.go/apps/api/openapi/gen.go (3)
464-521: LGTM!The new error response types (
ServiceUnavailableErrorResponse,TooManyRequestsErrorResponse, andUnprocessableEntityErrorResponse) are properly structured and consistent with existing error response types. The comprehensive documentation comments provide clear guidance on when each error occurs and how to handle it.
549-566: LGTM!The analytics verification types are well-designed. The use of
[]map[string]interface{}forV2AnalyticsGetVerificationsResponseDatais appropriate since the returned fields vary based on the SQL SELECT clause. The comments clearly explain this flexible behavior.
2083-2084: LGTM!The request body alias follows the established pattern used throughout the codebase for JSON content type request bodies.
go/apps/api/openapi/openapi-generated.yaml (2)
152-169: LGTM: NotFound error shape and semanticsSchema and description align with the repo’s NOT_FOUND-on-auth pattern to reduce information leakage.
6213-6214: LGTM: analytics tag addedGood categorization for the new surface.
There was a problem hiding this comment.
Actionable comments posted: 2
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
deployment/docker-compose.yaml(4 hunks)
🧰 Additional context used
🧠 Learnings (2)
📓 Common learnings
Learnt from: Flo4604
Repo: unkeyed/unkey PR: 4190
File: go/internal/services/keys/verifier.go:51-53
Timestamp: 2025-10-30T15:10:52.743Z
Learning: PR #4190 for unkeyed/unkey is focused solely on database schema and query changes for identity-based credits. It adds IdentityCredits and KeyCredits fields to structs and queries, but does not implement the priority enforcement logic in the usagelimiter. The logic implementation is intentionally deferred to a later PR in the stack.
📚 Learning: 2025-07-22T09:02:12.495Z
Learnt from: chronark
Repo: unkeyed/unkey PR: 3638
File: deployment/docker-compose.yaml:81-94
Timestamp: 2025-07-22T09:02:12.495Z
Learning: The docker-compose.yaml file in deployment/ is specifically for development environments, not production. Kafka and other service configurations in this file should be optimized for development convenience rather than production security/hardening.
Applied to files:
deployment/docker-compose.yaml
🪛 Gitleaks (8.28.0)
deployment/docker-compose.yaml
[high] 93-93: Detected a Generic API Key, potentially exposing access to various services and sensitive operations.
(generic-api-key)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (6)
- GitHub Check: Test Go API Local / Test
- GitHub Check: Build / Build
- GitHub Check: Test API / API Test Local
- GitHub Check: Test Dashboard / Test Dashboard
- GitHub Check: Test Packages / Test
- GitHub Check: Analyze (javascript-typescript)
🔇 Additional comments (3)
deployment/docker-compose.yaml (3)
74-75: Addition of s3 service health dependency looks good.The apiv2 service now depends on s3 being healthy, which aligns with the analytics/vault integration changes in this PR. No concerns with this change.
89-92: Environment variable naming is now consistent with UNKEY_ prefix convention.The renaming from
VAULT_S3_*toUNKEY_VAULT_S3_*improves consistency across the configuration. The newUNKEY_CLICKHOUSE_ANALYTICS_URLaligns with the analytics feature additions. This looks good, aside from the secret handling concern flagged separately.Also applies to: 95-95
89-95: No security issue: This is a local development file with shared example credentials.The secret value is not a real credential but an example value that appears identically across multiple development files (docker-compose.yaml, api.yaml, gw.yaml, ctrl.yaml), indicating it's an intentional test/placeholder. All surrounding environment variables use generic test credentials (password, test-token-123, minio_root_user, etc.), and the file is committed to git—not excluded by .gitignore. This is consistent with the repo's established practice of using hardcoded generic credentials in local development environments for convenience.
Likely an incorrect or invalid review comment.
| image: bufbuild/bufstream:0.4.4 | ||
| network_mode: host | ||
| command: ["serve", "--inmemory"] |
There was a problem hiding this comment.
🧩 Analysis chain
Critical: Kafka network_mode: host breaks service discovery for apiv2.
The change from the previous Kafka configuration to network_mode: host with the simplified Bufstream setup creates a networking problem. The apiv2 service (line 63) uses the default bridge network (networks: - default), but Kafka now operates in the host's network namespace. When apiv2 tries to connect to kafka:9092 (line 94), the DNS name kafka will not resolve because:
network_mode: hostbypasses Docker's internal DNS and container networking- apiv2 in the default network cannot reach services in host mode
This will cause connection failures. Either:
- Add Kafka back to the
networks: - defaultsection and expose ports, or - Change all dependent services (apiv2, etc.) to also use
network_mode: hostand update connection strings tolocalhost:9092
The previous configuration (despite the advertised listeners issue flagged in past reviews) did work for inter-container communication. Verify this change is intentional and test inter-service connectivity before merging.
🏁 Script executed:
#!/bin/bash
# Extract the relevant sections of the docker-compose.yaml file
echo "=== Lines 60-70 (apiv2 service) ==="
sed -n '60,70p' deployment/docker-compose.yaml
echo ""
echo "=== Lines 88-100 (Kafka service environment) ==="
sed -n '88,100p' deployment/docker-compose.yaml
echo ""
echo "=== Lines 114-120 (Kafka service image/network/command) ==="
sed -n '114,120p' deployment/docker-compose.yaml
echo ""
echo "=== Lines 450-465 (volumes section) ==="
sed -n '450,465p' deployment/docker-compose.yaml
echo ""
echo "=== Check if kafka service has networks defined ==="
sed -n '105,125p' deployment/docker-compose.yamlLength of output: 1872
🌐 Web query:
Docker Compose network_mode host service DNS resolution other containers
💡 Result:
Short answer: A service using network_mode: host does NOT join Docker Compose networks and therefore is not visible to Docker's embedded DNS — it cannot resolve other services by their Compose service names. Use one of these fixes:
- Don’t use host mode; use a user-defined bridge network so service names resolve. [1][2]
- Have the host-mode container talk to other services via published host ports (localhost:PORT) or the other containers' host IPs. [2]
- Add static entries (extra_hosts or /etc/hosts) or run an external DNS reachable from the host network and point the container's DNS to it. [1][3]
- If you must mix host-mode and bridge-mode, expose needed services on the host (publish ports) or use an external network/DNS service for name resolution. [1][3]
References: Docker Compose networking (user networks & service discovery). [1] network_mode: host behavior and caveats. [2] Community explanations showing host-mode containers can’t use Docker embedded DNS for other containers. [3]
Critical: Kafka network_mode: host breaks service discovery for apiv2.
The change to network_mode: host (line 116) creates a confirmed networking problem. Services using network_mode: host do not join Docker Compose networks and are therefore invisible to Docker's embedded DNS, making them unreachable by service name.
The apiv2 service (lines 61-62) uses the default bridge network and attempts to connect to UNKEY_KAFKA_BROKERS: "kafka:9092" (line 95). This DNS name will not resolve because Kafka operates in the host network namespace, not Docker's internal network.
This will cause connection failures. Choose one approach:
- Remove
network_mode: hostfrom Kafka and add it to a Docker network, or - Change all dependent services to also use
network_mode: hostand update connection strings tolocalhost:9092
Additional issues:
kafka_datavolume is defined (line 455) but not mounted in the Kafka service—either mount it or remove the definition- Line 91 contains a base64-encoded secret flagged by security scanning; verify this should be committed in the dev environment
🤖 Prompt for AI Agents
In deployment/docker-compose.yaml around lines 114-116, the Kafka service is set
to network_mode: host which prevents it from joining Docker Compose networks and
breaks DNS-based service discovery for apiv2; fix by removing network_mode: host
and attaching Kafka to the same compose network as apiv2 (or alternatively make
all dependent services use host networking and update broker addresses to
localhost:9092), ensure the kafka_data volume defined at line 455 is actually
mounted into the Kafka service or remove the unused volume definition, and
verify whether the base64 secret on line 91 should be committed in the dev repo
(rotate or move to env/secret manager if not appropriate).
| clickhouse: | ||
| clickhouse-keeper: | ||
| s3: | ||
| kafka_data: |
There was a problem hiding this comment.
Unused volume definition: kafka_data is defined but not mounted.
Line 459 adds kafka_data: to the volumes section, but the Kafka service (lines 112–117) does not mount this volume. Either:
- Add
volumes: - kafka_data:/path/to/datato the Kafka service if data persistence is intended, or - Remove the unused volume definition
With the current --inmemory command flag, persistent storage is not needed, so removing the unused volume is likely correct.
🤖 Prompt for AI Agents
In deployment/docker-compose.yaml around line 459, the volume kafka_data is
declared but never mounted by any service; either remove this unused volume
entry or attach it to the Kafka service if persistence is required. Given the
compose uses the --inmemory flag, delete the kafka_data: entry from the volumes
section to avoid dead configuration, or alternatively add a volumes: -
kafka_data:/var/lib/kafka (or your Kafka data path) under the Kafka service if
you intend to persist data.
3 participants
What does this PR do?
This does too many things, but mainly:
Extends our cache package to Many Operations (SWRMany GetMany SetMany) and so on, also adds a tests for this
Clickhouse user "management"
Seed analytics data but in go
2 New permissions for reading analytics api.*.read_analytics and api.<api_id>.read_analytics.
Error docs + analytics docs
New endpoint that lets the user query clickhouse with his user and with that
Adds a clickhouse query parser which does things
Regenerated the sql schema for our ts client
Regenerated our protobuf generation stuff
Fixes #3562
Type of change
How should this be tested?
(edited by @chronark)
make upThen,
pnpm localto trigger the build and generate new permissionsYou can use your favorite curl/httpie/gui and call the endpoint
So please go trough the docs and run the queries from there or write your owns and test if it works
Checklist
Required
pnpm buildpnpm fmtconsole.logsgit pull origin mainAppreciated