Skip to content

fix(backend): prevent duplicate index jobs from indexedAt race #1298

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
brendan-kellam merged 2 commits into from
Jun 10, 2026
Merged

fix(backend): prevent duplicate index jobs from indexedAt race #1298

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
74ba58a
fix(backend): prevent duplicate index jobs from indexedAt race
brendan-kellam Jun 10, 2026
6b2b9ea
docs: add CHANGELOG entry for #1298
brendan-kellam Jun 10, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
- Surfaced an actionable error when the Lighthouse licensing service is unreachable, instead of a generic "unexpected error". [#1293](https://github.com/sourcebot-dev/sourcebot/pull/1293)
- Fixed the selected language model rapidly flipping in local storage after a language model was removed. [#1295](https://github.com/sourcebot-dev/sourcebot/pull/1295)
- Fixed issue where using multiple identity providers of the same type (e.g., gitlab) would result in unexpected behaviours. [#1177](https://github.com/sourcebot-dev/sourcebot/pull/1177)
- Fixed a race condition where large repositories could be indexed twice within a single reindex interval. [#1298](https://github.com/sourcebot-dev/sourcebot/pull/1298)

## [5.0.1] - 2026-06-04

Expand Down
39 changes: 17 additions & 22 deletions packages/backend/src/repoIndexManager.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -525,17 +525,16 @@ describe('RepoIndexManager', () => {

manager = new RepoIndexManager(mockPrisma, mockSettings, mockRedis, mockPromClient as any);

// The onJobCompleted handler reads the job via findUniqueOrThrow, then marks it
// COMPLETED and updates the repo (indexedAt, etc.) in a single repoIndexingJob.update
// with a nested repo update.
(mockPrisma.repoIndexingJob.findUniqueOrThrow as Mock).mockResolvedValue({
status: RepoIndexingJobStatus.PENDING,
});
// The onJobCompleted handler calls repoIndexingJob.update once and then repo.update
(mockPrisma.repoIndexingJob.update as Mock).mockResolvedValue({
type: RepoIndexingJobType.INDEX,
repoId: repo.id,
repo,
metadata: {},
});
(mockPrisma.repo.update as Mock).mockResolvedValue(repo);
(mockPrisma.repoIndexingJob.update as Mock).mockResolvedValue({ repo });

// Get the onCompleted handler
const onCompletedHandler = mockWorkerOn.mock.calls.find((call: unknown[]) => call[0] === 'completed')?.[1];
Expand All @@ -552,22 +551,20 @@ describe('RepoIndexManager', () => {

await onCompletedHandler(mockJob);

// The job status and indexedAt must be written together (single transaction) to
// close the race where the scheduler sees a completed job but a stale indexedAt.
expect(mockPrisma.repoIndexingJob.update).toHaveBeenCalledWith(
expect.objectContaining({
where: { id: 'job-1' },
data: expect.objectContaining({
status: RepoIndexingJobStatus.COMPLETED,
completedAt: expect.any(Date),
}),
})
);

expect(mockPrisma.repo.update).toHaveBeenCalledWith(
expect.objectContaining({
where: { id: repo.id },
data: expect.objectContaining({
indexedAt: expect.any(Date),
indexedCommitHash: 'abc123',
repo: {
update: expect.objectContaining({
indexedAt: expect.any(Date),
indexedCommitHash: 'abc123',
}),
},
}),
})
);
Expand Down Expand Up @@ -754,14 +751,12 @@ describe('RepoIndexManager', () => {
manager = new RepoIndexManager(mockPrisma, mockSettings, mockRedis, mockPromClient as any);

(mockPrisma.repoIndexingJob.findUniqueOrThrow as Mock).mockResolvedValue({
status: RepoIndexingJobStatus.PENDING,
});
(mockPrisma.repoIndexingJob.update as Mock).mockResolvedValue({
type: RepoIndexingJobType.CLEANUP,
repoId: repo.id,
repo,
metadata: {},
});
(mockPrisma.repoIndexingJob.update as Mock).mockResolvedValue({ repo });
(mockPrisma.repo.delete as Mock).mockResolvedValue(repo);

const onCompletedHandler = mockWorkerOn.mock.calls.find((call: unknown[]) => call[0] === 'completed')?.[1];
Expand Down Expand Up @@ -840,13 +835,13 @@ describe('RepoIndexManager', () => {

manager = new RepoIndexManager(mockPrisma, mockSettings, mockRedis, mockPromClient as any);

(mockPrisma.repoIndexingJob.update as Mock).mockResolvedValue({
(mockPrisma.repoIndexingJob.findUniqueOrThrow as Mock).mockResolvedValue({
type: RepoIndexingJobType.INDEX,
repoId: repo.id,
repo,
metadata: {},
});
(mockPrisma.repo.update as Mock).mockResolvedValue(repo);
(mockPrisma.repoIndexingJob.update as Mock).mockResolvedValue({ repo });

const onCompletedHandler = mockWorkerOn.mock.calls.find((call: unknown[]) => call[0] === 'completed')?.[1];

Expand All @@ -867,9 +862,9 @@ describe('RepoIndexManager', () => {
data: expect.objectContaining({
status: RepoIndexingJobStatus.COMPLETED,
repo: {
update: {
update: expect.objectContaining({
latestIndexingJobStatus: RepoIndexingJobStatus.COMPLETED,
},
}),
},
}),
})
Expand Down
65 changes: 38 additions & 27 deletions packages/backend/src/repoIndexManager.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -545,23 +545,16 @@ export class RepoIndexManager {
private async onJobCompleted(job: Job<JobPayload>) {
try {
const logger = createJobLogger(job.data.jobId);
const jobData = await this.db.repoIndexingJob.update({
const jobData = await this.db.repoIndexingJob.findUniqueOrThrow({
where: { id: job.data.jobId },
data: {
status: RepoIndexingJobStatus.COMPLETED,
completedAt: new Date(),
repo: {
update: {
latestIndexingJobStatus: RepoIndexingJobStatus.COMPLETED,
}
}
},
include: {
repo: true,
}
});

const jobTypeLabel = getJobTypePrometheusLabel(jobData.type);
// @note: capture this before the update below, since the update sets indexedAt.
const isFirstIndex = jobData.repo.indexedAt === null;

if (jobData.type === RepoIndexingJobType.INDEX) {
const { path: repoPath } = getRepoPath(jobData.repo);
Expand All @@ -576,29 +569,47 @@ export class RepoIndexManager {

const jobMetadata = repoIndexingJobMetadataSchema.parse(jobData.metadata);

const repo = await this.db.repo.update({
where: { id: jobData.repoId },
const { repo } = await this.db.repoIndexingJob.update({
where: { id: job.data.jobId },
data: {
indexedAt: new Date(),
indexedCommitHash: commitHash,
pushedAt: pushedAt,
metadata: {
...(jobData.repo.metadata as RepoMetadata),
indexedRevisions: jobMetadata.indexedRevisions,
} satisfies RepoMetadata,
// @note: always update the default branch. While this field can be set
// during connection syncing, by setting it here we ensure that a) the
// default branch is as up to date as possible (since repo indexing happens
// more frequently than connection syncing) and b) for hosts where it is
// impossible to determine the default branch from the host's API
// (e.g., generic git url), we still set the default branch here.
defaultBranch: defaultBranch,
status: RepoIndexingJobStatus.COMPLETED,
completedAt: new Date(),
repo: {
update: {
latestIndexingJobStatus: RepoIndexingJobStatus.COMPLETED,
indexedAt: new Date(),
indexedCommitHash: commitHash,
pushedAt: pushedAt,
metadata: {
...(jobData.repo.metadata as RepoMetadata),
indexedRevisions: jobMetadata.indexedRevisions,
} satisfies RepoMetadata,
// @note: always update the default branch. While this field can be set
// during connection syncing, by setting it here we ensure that a) the
// default branch is as up to date as possible (since repo indexing happens
// more frequently than connection syncing) and b) for hosts where it is
// impossible to determine the default branch from the host's API
// (e.g., generic git url), we still set the default branch here.
defaultBranch: defaultBranch,
}
}
},
include: {
repo: true,
}
});

logger.debug(`Completed index job ${job.data.jobId} for repo ${repo.name} (id: ${repo.id})`);
}
else if (jobData.type === RepoIndexingJobType.CLEANUP) {
await this.db.repoIndexingJob.update({
where: { id: job.data.jobId },
data: {
status: RepoIndexingJobStatus.COMPLETED,
completedAt: new Date(),
}
});

const repo = await this.db.repo.delete({
where: { id: jobData.repoId },
});
Expand All @@ -610,7 +621,7 @@ export class RepoIndexManager {
this.promClient.activeRepoIndexJobs.dec({ repo: job.data.repoName, type: jobTypeLabel });
this.promClient.repoIndexJobSuccessTotal.inc({ repo: job.data.repoName, type: jobTypeLabel });

if (jobData.type === RepoIndexingJobType.INDEX && jobData.repo.indexedAt === null) {
if (jobData.type === RepoIndexingJobType.INDEX && isFirstIndex) {
captureEvent('backend_repo_first_indexed', {
repoId: job.data.repoId,
type: jobData.repo.external_codeHostType,
Expand Down
Loading