Random fixes and improvements (#244)

.env.development

@@ -34,7 +34,7 @@ AUTH_URL="http://localhost:3000"
 # Sentry
 # SENTRY_BACKEND_DSN=""
 # NEXT_PUBLIC_SENTRY_WEBAPP_DSN=""
-# SENTRY_ENVIRONMENT="dev"
+SENTRY_ENVIRONMENT="dev"
 # NEXT_PUBLIC_SENTRY_ENVIRONMENT="dev"
 # SENTRY_AUTH_TOKEN=
 

packages/backend/src/connectionManager.ts

@@ -1,4 +1,4 @@
-import { Connection, ConnectionSyncStatus, PrismaClient, Prisma } from "@sourcebot/db";
+import { Connection, ConnectionSyncStatus, PrismaClient, Prisma, RepoIndexingStatus } from "@sourcebot/db";
 import { Job, Queue, Worker } from 'bullmq';
 import { Settings } from "./types.js";
 import { ConnectionConfig } from "@sourcebot/schemas/v3/connection.type";
@@ -160,7 +160,7 @@ export class ConnectionManager implements IConnectionManager {
             }
         }
 
-        const { repoData, notFound } = result;
+        let { repoData, notFound } = result;
 
         // Push the information regarding not found users, orgs, and repos to the connection's syncStatusMetadata. Note that 
         // this won't be overwritten even if the connection job fails
@@ -174,7 +174,7 @@ export class ConnectionManager implements IConnectionManager {
         });
             
         // Filter out any duplicates by external_id and external_codeHostUrl.
-        repoData.filter((repo, index, self) => {
+        repoData = repoData.filter((repo, index, self) => {
             return index === self.findIndex(r =>
                 r.external_id === repo.external_id &&
                 r.external_codeHostUrl === repo.external_codeHostUrl
@@ -263,6 +263,14 @@ export class ConnectionManager implements IConnectionManager {
 
     private async onSyncJobFailed(job: Job | undefined, err: unknown) {
         this.logger.info(`Connection sync job failed with error: ${err}`);
+        Sentry.captureException(err, {
+            tags: {
+                repoId: job?.data.repo.id,
+                jobId: job?.id,
+                queue: QUEUE_NAME,
+            }
+        });
+
         if (job) {
             const { connectionId } = job.data;
 

packages/backend/src/repoManager.ts

@@ -130,15 +130,25 @@ export class RepoManager implements IRepoManager {
         const thresholdDate = new Date(Date.now() - this.settings.reindexIntervalMs);
         const repos = await this.db.repo.findMany({
             where: {
-                repoIndexingStatus: {
-                    in: [
-                        RepoIndexingStatus.NEW,
-                        RepoIndexingStatus.INDEXED
-                    ]
-                },
                 OR: [
-                    { indexedAt: null },
-                    { indexedAt: { lt: thresholdDate } },
+                    // "NEW" is really a misnomer here - it just means that the repo needs to be indexed
+                    // immediately. In most cases, this will be because the repo was just created and
+                    // is indeed "new". However, it could also be that a "retry" was requested on a failed
+                    // index. So, we don't want to block on the indexedAt timestamp here.
+                    {
+                        repoIndexingStatus: RepoIndexingStatus.NEW,
+                    },
+                    // When the repo has already been indexed, we only want to reindex if the reindexing
+                    // interval has elapsed (or if the date isn't set for some reason).
+                    {
+                        AND: [
+                            { repoIndexingStatus: RepoIndexingStatus.INDEXED },
+                            { OR: [
+                                { indexedAt: null },
+                                { indexedAt: { lt: thresholdDate } },
+                            ]}
+                        ]
+                    }
                 ]
             },
             include: {
@@ -335,7 +345,15 @@ export class RepoManager implements IRepoManager {
     }
 
     private async onIndexJobFailed(job: Job<RepoIndexingPayload> | undefined, err: unknown) {
-        this.logger.info(`Repo index job failed (id: ${job?.id ?? 'unknown'})`);
+        this.logger.info(`Repo index job failed (id: ${job?.id ?? 'unknown'}) with error: ${err}`);
+        Sentry.captureException(err, {
+            tags: {
+                repoId: job?.data.repo.id,
+                jobId: job?.id,
+                queue: REPO_INDEXING_QUEUE,
+            }
+        });
+
         if (job) {
             this.promClient.activeRepoIndexingJobs.dec();
             this.promClient.repoIndexingFailTotal.inc();
@@ -474,6 +492,13 @@ export class RepoManager implements IRepoManager {
 
     private async onGarbageCollectionJobFailed(job: Job<RepoGarbageCollectionPayload> | undefined, err: unknown) {
         this.logger.info(`Garbage collection job failed (id: ${job?.id ?? 'unknown'}) with error: ${err}`);
+        Sentry.captureException(err, {
+            tags: {
+                repoId: job?.data.repo.id,
+                jobId: job?.id,
+                queue: REPO_GC_QUEUE,
+            }
+        });
 
         if (job) {
             this.promClient.activeRepoGarbageCollectionJobs.dec();

packages/web/src/app/posthogProvider.tsx

@@ -38,8 +38,8 @@ export function PostHogProvider({ children, disabled }: PostHogProviderProps) {
                 // @see next.config.mjs for path rewrites to the "/ingest" route.
                 api_host: "/ingest",
                 person_profiles: 'identified_only',
-                capture_pageview: false, // @nocheckin Disable automatic pageview capture if we're not in public demo mode
-                autocapture: false, // Disable automatic event capture
+                capture_pageview: false,
+                autocapture: false,
                 // eslint-disable-next-line @typescript-eslint/no-explicit-any
                 sanitize_properties: (properties: Record<string, any>, _event: string) => {
                     // https://posthog.com/docs/libraries/js#config

packages/web/src/initialize.ts

@@ -1,4 +1,4 @@
-import { ConnectionSyncStatus, OrgRole, Prisma } from '@sourcebot/db';
+import { ConnectionSyncStatus, OrgRole, Prisma, RepoIndexingStatus } from '@sourcebot/db';
 import { env } from './env.mjs';
 import { prisma } from "@/prisma";
 import { SINGLE_TENANT_USER_ID, SINGLE_TENANT_ORG_ID, SINGLE_TENANT_ORG_DOMAIN, SINGLE_TENANT_ORG_NAME, SINGLE_TENANT_USER_EMAIL } from './lib/constants';
@@ -105,6 +105,13 @@ const initSingleTenancy = async () => {
                             name: key,
                             orgId: SINGLE_TENANT_ORG_ID,
                         }
+                    },
+                    include: {
+                        repos: {
+                            include: {
+                                repo: true,
+                            }
+                        }
                     }
                 });
 
@@ -137,6 +144,21 @@ const initSingleTenancy = async () => {
                 });
 
                 console.log(`Upserted connection with name '${key}'. Connection ID: ${connectionDb.id}`);
+
+                // Re-try any repos that failed to index.
+                const failedRepos = currentConnection?.repos.filter(repo => repo.repo.repoIndexingStatus === RepoIndexingStatus.FAILED).map(repo => repo.repo.id) ?? [];
+                if (failedRepos.length > 0) {
+                    await prisma.repo.updateMany({
+                        where: {
+                            id: {
+                                in: failedRepos,
+                            }
+                        },
+                        data: {
+                            repoIndexingStatus: RepoIndexingStatus.NEW,
+                        }
+                    })
+                }
             }
         }
     }