Skip to content

Job application tracker and AI-powered job search assistant. Helps job seekers manage their search journey with AI resume review, job matching, task logging, and application analytics, all while keeping your data private.

feat: add partial unique indexes to prevent duplicate automation runs and discovered jobs

Khuram Niaz3d13ac8bf000
diff --git a/prisma/migrations/20260710000001_automation_run_single_active/migration.sql b/prisma/migrations/20260710000001_automation_run_single_active/migration.sql
new file mode 100644
index 0000000..338c039
--- /dev/null
+++ b/prisma/migrations/20260710000001_automation_run_single_active/migration.sql
@@ -0,0 +1,12 @@
+-- Prevents two AutomationRun rows from being "running"/"cancelling" at once
+-- for the same automation. The prior guard (SELECT-then-INSERT inside a
+-- $transaction) assumed SQLite serializes concurrent write transactions,
+-- but deferred transactions don't take a write lock until the first write,
+-- so two concurrent runAutomation() calls could both pass the SELECT check
+-- before either INSERTed, producing duplicate concurrent runs (and, in turn,
+-- duplicate discovered jobs since each run computed its own dedup snapshot).
+-- Partial index (not expressible via Prisma's schema DSL) so historical
+-- completed/failed/cancelled runs are unaffected.
+CREATE UNIQUE INDEX "AutomationRun_automationId_active_key"
+ON "AutomationRun"("automationId")
+WHERE "status" IN ('running', 'cancelling');
diff --git a/prisma/migrations/20260710000002_job_automation_url_unique/migration.sql b/prisma/migrations/20260710000002_job_automation_url_unique/migration.sql
new file mode 100644
index 0000000..6e76ce4
--- /dev/null
+++ b/prisma/migrations/20260710000002_job_automation_url_unique/migration.sql
@@ -0,0 +1,9 @@
+-- Guards against duplicate automation-discovered jobs at the DB level, since
+-- app-level dedup (dedupeJobs in scraper/utils.ts) reads existing URLs at the
+-- start of a run and can't see another concurrent run's in-flight saves.
+-- Scoped to automationId IS NOT NULL so it never affects manually-created
+-- jobs, which legitimately reuse placeholder/generic URLs. Partial index,
+-- not expressible via Prisma's schema DSL.
+CREATE UNIQUE INDEX "Job_userId_jobUrl_automation_key"
+ON "Job"("userId", "jobUrl")
+WHERE "automationId" IS NOT NULL;
diff --git a/prisma/schema.prisma b/prisma/schema.prisma
index 3758aa8..1744d9f 100644
--- a/prisma/schema.prisma
+++ b/prisma/schema.prisma
@@ -325,6 +325,9 @@ model Job {
@@index([userId, automationId])
@@index([userId, discoveryStatus])
+ // Also has a hand-written partial unique index (not expressible here):
+ // UNIQUE(userId, jobUrl) WHERE automationId IS NOT NULL
+ // See migrations/20260710000002_job_automation_url_unique
}
model ActivityType {
@@ -439,6 +442,9 @@ model AutomationRun {
@@index([automationId])
@@index([startedAt])
+ // Also has a hand-written partial unique index (not expressible here):
+ // UNIQUE(automationId) WHERE status IN ('running', 'cancelling')
+ // See migrations/20260710000001_automation_run_single_active
}
model Note {
diff --git a/src/app/api/automations/[id]/run/route.ts b/src/app/api/automations/[id]/run/route.ts
index 844d087..b63b4ec 100644
--- a/src/app/api/automations/[id]/run/route.ts
+++ b/src/app/api/automations/[id]/run/route.ts
@@ -3,7 +3,11 @@ import "server-only";
import { NextRequest, NextResponse } from "next/server";
import { auth } from "@/auth";
import db from "@/lib/db";
-import { runAutomation, getUserAiSettings } from "@/lib/scraper";
+import {
+ runAutomation,
+ getUserAiSettings,
+ AutomationAlreadyRunningError,
+} from "@/lib/scraper";
import { PROVIDER_VERIFIERS } from "@/lib/ai/provider-registry.server";
import { getOllamaBaseUrl } from "@/actions/apiKey.actions";
import { AiProvider } from "@/models/ai.model";
@@ -127,6 +131,10 @@ export async function POST(
createdAt: automation.createdAt,
updatedAt: automation.updatedAt,
}).catch((err) => {
+ if (err instanceof AutomationAlreadyRunningError) {
+ console.log(`Skipping manual run for ${automation.id} - run already in progress`);
+ return;
+ }
console.error("Background automation run failed:", err);
});
diff --git a/src/components/automations/DiscoveredJobDetail.tsx b/src/components/automations/DiscoveredJobDetail.tsx
index 8fa6b6f..39b6c49 100644
--- a/src/components/automations/DiscoveredJobDetail.tsx
+++ b/src/components/automations/DiscoveredJobDetail.tsx
@@ -16,7 +16,6 @@ import { toast } from "@/components/ui/use-toast";
import {
Building2,
MapPin,
- ExternalLink,
Check,
X,
Loader2,
@@ -25,6 +24,7 @@ import {
import type { DiscoveredJob } from "@/models/automation.model";
import type { JobMatchData } from "@/models/ai.schemas";
import { getWorkplaceTypeLabel } from "@/models/job.model";
+import { formatUrl } from "@/lib/utils";
import {
acceptDiscoveredJob,
dismissDiscoveredJob,
@@ -130,19 +130,7 @@ export function DiscoveredJobDetail({
<DialogHeader>
<div className="flex items-start justify-between gap-4 pr-8">
<div className="min-w-0 space-y-1.5">
- <DialogTitle className="flex items-center gap-2">
- {job.JobTitle.label}
- {job.jobUrl && (
- <a
- href={job.jobUrl}
- target="_blank"
- rel="noopener noreferrer"
- className="shrink-0 text-muted-foreground hover:text-foreground"
- >
- <ExternalLink className="h-4 w-4" />
- </a>
- )}
- </DialogTitle>
+ <DialogTitle>{job.JobTitle.label}</DialogTitle>
<DialogDescription className="flex flex-wrap items-start gap-x-4 gap-y-1">
<span className="flex items-start gap-1">
<Building2 className="h-4 w-4 shrink-0 mt-0.5" />
@@ -186,6 +174,19 @@ export function DiscoveredJobDetail({
<ScrollArea className="max-h-[60vh]">
<div className="space-y-4 pr-4">
+ {job.jobUrl && (
+ <div>
+ <span className="font-semibold mr-2">Job URL:</span>
+ <a
+ href={formatUrl(job.jobUrl)}
+ target="_blank"
+ rel="noopener noreferrer"
+ >
+ {job.jobUrl}
+ </a>
+ </div>
+ )}
+
<div className="flex items-center gap-4">
<Badge variant="outline">{job.discoveryStatus}</Badge>
{job.automation && (
diff --git a/src/lib/scheduler/index.ts b/src/lib/scheduler/index.ts
index c930a73..5edc5fc 100644
--- a/src/lib/scheduler/index.ts
+++ b/src/lib/scheduler/index.ts
@@ -1,7 +1,7 @@
import cron, { ScheduledTask } from "node-cron";
import { SCHEDULER_CONSTANTS } from "@/lib/constants";
import db from "@/lib/db";
-import { runAutomation } from "@/lib/scraper";
+import { runAutomation, AutomationAlreadyRunningError } from "@/lib/scraper";
import type { JobBoard } from "@/models/automation.model";
let scheduledTask: ScheduledTask | null = null;
@@ -78,6 +78,10 @@ async function runDueAutomations() {
});
console.log(`[Scheduler] Automation ${automation.name} completed: ${result.status}, saved ${result.jobsSaved} jobs`);
} catch (error) {
+ if (error instanceof AutomationAlreadyRunningError) {
+ console.log(`[Scheduler] Skipping automation ${automation.id} - run already in progress`);
+ continue;
+ }
const message = error instanceof Error ? error.message : "Unknown error";
console.error(`[Scheduler] Automation ${automation.name} failed:`, message);
}
diff --git a/src/lib/scraper/index.ts b/src/lib/scraper/index.ts
index aaeae3c..c111cf9 100644
--- a/src/lib/scraper/index.ts
+++ b/src/lib/scraper/index.ts
@@ -23,5 +23,6 @@ export {
export {
runAutomation,
getUserAiSettings,
+ AutomationAlreadyRunningError,
type RunnerResult,
} from "./runner";
diff --git a/src/lib/scraper/runner.ts b/src/lib/scraper/runner.ts
index 25fa7fb..b73b0a5 100644
--- a/src/lib/scraper/runner.ts
+++ b/src/lib/scraper/runner.ts
@@ -153,20 +153,47 @@ interface ResumeWithSections extends PrismaResume {
}>;
}
+// Thrown by runAutomation's atomic claim when another run is already active
+// for this automation. Callers (scheduler, manual-run route) already do their
+// own pre-check for a fast/clean skip, but that check-then-act isn't atomic on
+// its own — two near-simultaneous callers can both pass it before either has
+// written a "running" row. This error signals the loser of that race so it
+// can skip instead of starting a duplicate run.
+export class AutomationAlreadyRunningError extends Error {
+ constructor(automationId: string) {
+ super(`Automation ${automationId} already has an active run`);
+ this.name = "AutomationAlreadyRunningError";
+ }
+}
+
export async function runAutomation(
automation: Automation,
signal?: AbortSignal,
): Promise<RunnerResult> {
+ // Atomic claim: a partial unique index (AutomationRun_automationId_active_key,
+ // see migrations/20260710000001_automation_run_single_active) allows at most
+ // one running/cancelling row per automation. A prior SELECT-then-INSERT
+ // $transaction assumed SQLite serializes concurrent write transactions,
+ // but deferred transactions don't take a write lock until the first write,
+ // so two racing callers could both pass the SELECT before either INSERTed —
+ // relying on the DB constraint instead closes that race for good.
+ let run;
+ try {
+ run = await db.automationRun.create({
+ data: {
+ automationId: automation.id,
+ status: "running",
+ },
+ });
+ } catch (error: any) {
+ if (error?.code === "P2002") {
+ throw new AutomationAlreadyRunningError(automation.id);
+ }
+ throw error;
+ }
+
console.log(`[Automation ${automation.id}] Starting automation run`);
automationLogger.startRun(automation.id);
-
- const run = await db.automationRun.create({
- data: {
- automationId: automation.id,
- status: "running",
- },
- });
-
console.log(`[Automation ${automation.id}] Created run with ID: ${run.id}`);
automationLogger.log(
automation.id,
@@ -517,7 +544,16 @@ export async function runAutomation(
`Job saved successfully (${jobsSaved} total)`,
{ jobsSaved },
);
- } catch (err) {
+ } catch (err: any) {
+ if (err?.code === "P2002") {
+ // Another concurrent run already saved this URL first.
+ automationLogger.log(
+ automation.id,
+ "info",
+ "Job skipped - already saved by a concurrent run",
+ );
+ return;
+ }
const errorMsg = err instanceof Error ? err.message : "Unknown error";
automationLogger.log(
automation.id,
@@ -696,12 +732,17 @@ function scalePrerank(raw: number): number {
return Math.min(99, Math.max(0, Math.round((raw / PRERANK_MAX) * 99)));
}
+// Returns false (instead of throwing) when a concurrent run already saved
+// this exact URL first — the Job_userId_jobUrl_automation_key partial unique
+// index (migrations/20260710000002_job_automation_url_unique) is the
+// backstop for that race, since app-level dedup only sees a point-in-time
+// snapshot of existing URLs.
async function persistDiscoveredJob(
automation: Automation,
job: JobDetails,
matchScore: number,
matchData: object,
-): Promise<void> {
+): Promise<boolean> {
const scrapedJob: ScrapedJobData = {
title: job.title,
company: job.company,
@@ -722,7 +763,13 @@ async function persistDiscoveredJob(
matchData: JSON.stringify(matchData),
});
- await db.job.create({ data: jobRecord });
+ try {
+ await db.job.create({ data: jobRecord });
+ return true;
+ } catch (err: any) {
+ if (err?.code === "P2002") return false;
+ throw err;
+ }
}
async function runAtsRun(
@@ -903,7 +950,7 @@ async function runAtsRun(
for (const scored of pipeline.toSaveUnanalyzed) {
if (signal?.aborted) break;
try {
- await persistDiscoveredJob(
+ const saved = await persistDiscoveredJob(
automation,
scored.job,
scalePrerank(scored.score),
@@ -913,7 +960,7 @@ async function runAtsRun(
analyzed: false,
},
);
- jobsSaved++;
+ if (saved) jobsSaved++;
} catch (err) {
console.error(`${label} Failed to save listing:`, err);
}
@@ -935,7 +982,7 @@ async function runAtsRun(
const saveUnanalyzed = async () => {
try {
- await persistDiscoveredJob(
+ const saved = await persistDiscoveredJob(
automation,
scored.job,
scalePrerank(scored.score),
@@ -945,7 +992,7 @@ async function runAtsRun(
analyzed: false,
},
);
- jobsSaved++;
+ if (saved) jobsSaved++;
} catch (err) {
console.error(`${label} Failed to save listing:`, err);
}
@@ -1007,18 +1054,23 @@ async function runAtsRun(
);
try {
- await persistDiscoveredJob(automation, scored.job, matchResult.score, {
- ...matchResult.data,
- resumeId: resume.id,
- resumeTitle: resume.title,
- matchedAt: new Date().toISOString(),
- provider: aiSettings.provider,
- model: modelName,
- prerankScore: scored.score,
- prerankComponents: scored.components,
- analyzed: true,
- });
- jobsSaved++;
+ const saved = await persistDiscoveredJob(
+ automation,
+ scored.job,
+ matchResult.score,
+ {
+ ...matchResult.data,
+ resumeId: resume.id,
+ resumeTitle: resume.title,
+ matchedAt: new Date().toISOString(),
+ provider: aiSettings.provider,
+ model: modelName,
+ prerankScore: scored.score,
+ prerankComponents: scored.components,
+ analyzed: true,
+ },
+ );
+ if (saved) jobsSaved++;
} catch (err) {
console.error(`${label} Failed to save analyzed job:`, err);
}