Add repo index validation (#339)

* add repo index validation

* add entry to changelog

Author: msukkari

CHANGELOG.md

@@ -11,6 +11,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Added seperate page for signup. [#311](https://github.com/sourcebot-dev/sourcebot/pull/331)
 - Fix repo images in authed instance case and add manifest json. [#332](https://github.com/sourcebot-dev/sourcebot/pull/332)
 - Added encryption logic for license keys. [#335](https://github.com/sourcebot-dev/sourcebot/pull/335)
+- Added repo shard validation on startup. [#339](https://github.com/sourcebot-dev/sourcebot/pull/339)
 - Added support for a file explorer when browsing files. [#336](https://github.com/sourcebot-dev/sourcebot/pull/336)
 
 ## [4.1.1] - 2025-06-03

packages/backend/src/main.ts

@@ -68,5 +68,6 @@ export const main = async (db: PrismaClient, context: AppContext) => {
     connectionManager.registerPollingCallback();
 
     const repoManager = new RepoManager(db, settings, redis, promClient, context);
+    await repoManager.validateIndexedReposHaveShards();
     await repoManager.blockingPollLoop();
 }

packages/backend/src/repoManager.ts

@@ -12,6 +12,7 @@ import { PromClient } from './promClient.js';
 import * as Sentry from "@sentry/node";
 
 interface IRepoManager {
+    validateIndexedReposHaveShards: () => Promise<void>;
     blockingPollLoop: () => void;
     dispose: () => void;
 }
@@ -526,6 +527,61 @@ export class RepoManager implements IRepoManager {
         }
     }
 
+    ///////////////////////////
+    // Repo index validation
+    ///////////////////////////
+
+    public async validateIndexedReposHaveShards() {
+        logger.info('Validating indexed repos have shards...');
+        
+        const indexedRepos = await this.db.repo.findMany({
+            where: {
+                repoIndexingStatus: RepoIndexingStatus.INDEXED
+            }
+        });
+        logger.info(`Found ${indexedRepos.length} repos in the DB marked as INDEXED`);
+
+        if (indexedRepos.length === 0) {
+            return;
+        }
+
+        const reposToReindex: number[] = [];
+
+        for (const repo of indexedRepos) {
+            const shardPrefix = getShardPrefix(repo.orgId, repo.id);
+            
+            // TODO: this doesn't take into account if a repo has multiple shards and only some of them are missing. To support that, this logic
+            // would need to know how many total shards are expected for this repo
+            let hasShards = false;
+            try {
+                const files = readdirSync(this.ctx.indexPath);
+                hasShards = files.some(file => file.startsWith(shardPrefix));
+            } catch (error) {
+                logger.error(`Failed to read index directory ${this.ctx.indexPath}: ${error}`);
+                continue;
+            }
+
+            if (!hasShards) {
+                logger.info(`Repo ${repo.displayName} (id: ${repo.id}) is marked as INDEXED but has no shards on disk. Marking for reindexing.`);
+                reposToReindex.push(repo.id);
+            }
+        }
+
+        if (reposToReindex.length > 0) {
+            await this.db.repo.updateMany({
+                where: {
+                    id: { in: reposToReindex }
+                },
+                data: {
+                    repoIndexingStatus: RepoIndexingStatus.NEW
+                }
+            });
+            logger.info(`Marked ${reposToReindex.length} repos for reindexing due to missing shards`);
+        }
+
+        logger.info('Done validating indexed repos have shards');
+    }
+
     private async fetchAndScheduleRepoTimeouts() {
         const repos = await this.db.repo.findMany({
             where: {

packages/web/src/initialize.ts

@@ -214,10 +214,8 @@ const initSingleTenancy = async () => {
     // Load any connections defined declaratively in the config file.
     const configPath = env.CONFIG_PATH;
     if (configPath) {
-        await syncDeclarativeConfig(configPath);
-
         // If we're given a config file, mark the org as onboarded so we don't go through
-        // the UI conneciton onboarding flow
+        // the UI connection onboarding flow
         await prisma.org.update({
             where: {
                 id: SINGLE_TENANT_ORG_ID,
@@ -226,7 +224,9 @@ const initSingleTenancy = async () => {
                 isOnboarded: true,
             }
         });
-
+        
+        await syncDeclarativeConfig(configPath);
+        
         // watch for changes assuming it is a local file
         if (!isRemotePath(configPath)) {
             watch(configPath, () => {