fix: comments

Signed-off-by: Umberto Sgueglia <usgueglia@contractor.linuxfoundation.org>

Author: ulemons

services/apps/automatic_onboarding_worker/src/activities/activities.ts

@@ -24,14 +24,21 @@ export async function fetchProjectsPendingOnboarding(
   return projects
 }
 
+async function findAlreadyOnboarded(
+  qx: ReturnType<typeof pgpQx>,
+  projectId: string,
+): Promise<IDbProjectCatalog | null> {
+  const fresh = await findProjectCatalogById(qx, projectId)
+  return fresh?.onboardedAt ? fresh : null
+}
+
 export async function onboardAndUpdateProject(project: IDbProjectCatalog): Promise<void> {
   const qx = pgpQx(svc.postgres.writer.connection())
   const startTime = Date.now()
 
-  // Guard: fetch fresh state to ensure the API is called at most once per project.
-  // Uses the writer connection to avoid replica lag missing a just-written onboardedAt.
-  const fresh = await findProjectCatalogById(qx, project.id)
-  if (fresh?.onboardedAt) {
+  // Guard: uses the writer connection to avoid replica lag missing a just-written onboardedAt.
+  const fresh = await findAlreadyOnboarded(qx, project.id)
+  if (fresh) {
     log.info(
       { id: project.id, repoUrl: project.repoUrl, onboardedAt: fresh.onboardedAt },
       'Project already onboarded, skipping API call.',
@@ -70,6 +77,16 @@ export async function markProjectOnboardingFailed(
 ): Promise<void> {
   const qx = pgpQx(svc.postgres.writer.connection())
 
+  // Guard: the write may have succeeded after the API call before retries were exhausted.
+  const fresh = await findAlreadyOnboarded(qx, projectId)
+  if (fresh) {
+    log.info(
+      { id: projectId, onboardedAt: fresh.onboardedAt },
+      'Project was already onboarded despite the reported failure, not marking as error.',
+    )
+    return
+  }
+
   await updateProjectCatalog(qx, projectId, {
     action: 'error',
     onboardingError: reason,

services/apps/automatic_onboarding_worker/src/schedules/scheduleProjectsOnboarding.ts

@@ -19,15 +19,14 @@ export const scheduleProjectsOnboarding = async () => {
       },
       policies: {
         overlap: ScheduleOverlapPolicy.SKIP,
-        catchupWindow: '1 minute',
+        catchupWindow: '1 hour',
       },
       action: {
         type: 'startWorkflow',
         workflowType: onboardProjects,
         taskQueue: 'automatic-onboarding',
         args: [ONBOARDING_ARGS],
-        // 50 projects × up to ~2min per attempt, up to 2 attempts each = ~3.3h worst case; set ceiling with margin.
-        workflowExecutionTimeout: '6 hours',
+        workflowExecutionTimeout: '14 hours',
         retry: {
           initialInterval: '30 seconds',
           backoffCoefficient: 2,

services/apps/automatic_onboarding_worker/src/workflows/onboardProjects.ts

@@ -9,8 +9,14 @@ const fetchActivities = proxyActivities<typeof activities>({
   retry: { maximumAttempts: 3 },
 })
 
-// Each onboarding call chains a segment create/query plus GitHub enrichment and integration calls; give generous headroom per project.
+// Each onboarding call chains a segment create/query plus GitHub enrichment and integration calls,
+// each of which can individually approach a ~30s backend timeout; give generous headroom per project.
 const onboardActivities = proxyActivities<typeof activities>({
+  startToCloseTimeout: '5 minutes',
+  retry: { maximumAttempts: 2 },
+})
+
+const failureActivities = proxyActivities<typeof activities>({
   startToCloseTimeout: '2 minutes',
   retry: { maximumAttempts: 2 },
 })
@@ -49,7 +55,7 @@ export async function onboardProjects(input: IOnboardProjectsInput = {}): Promis
       )
 
       try {
-        await onboardActivities.markProjectOnboardingFailed(project.id, reason)
+        await failureActivities.markProjectOnboardingFailed(project.id, reason)
       } catch (markErr) {
         // Don't let a failure to record the error state abort the rest of the batch.
         log.error(`Failed to mark project id=${project.id} as errored: ${String(markErr)}`)

services/libs/data-access-layer/src/project-catalog/projectCatalog.ts

@@ -336,7 +336,7 @@ export async function upsertProjectCatalog(
       "repoName" = EXCLUDED."repoName",
       "source" = COALESCE(EXCLUDED."source", "projectCatalog"."source"),
       "action" = CASE
-        WHEN "projectCatalog"."action" IN ('onboard', 'skip', 'unsure') THEN "projectCatalog"."action"
+        WHEN "projectCatalog"."action" IN ('onboard', 'skip', 'unsure', 'error') THEN "projectCatalog"."action"
         WHEN EXCLUDED.action = 'evaluate' THEN 'evaluate'
         ELSE "projectCatalog"."action"
       END,
@@ -409,7 +409,7 @@ export async function bulkUpsertProjectCatalog(
       "repoName" = EXCLUDED."repoName",
       "source" = COALESCE(EXCLUDED."source", "projectCatalog"."source"),
       "action" = CASE
-        WHEN "projectCatalog"."action" IN ('onboard', 'skip', 'unsure') THEN "projectCatalog"."action"
+        WHEN "projectCatalog"."action" IN ('onboard', 'skip', 'unsure', 'error') THEN "projectCatalog"."action"
         WHEN EXCLUDED.action = 'evaluate' THEN 'evaluate'
         ELSE "projectCatalog"."action"
       END,