Get startedDashboard

Nobody should be on-call

Polylane finds issues in your app and fixes them for you, automatically.

Trusted by the teams you trust

Polylane drives from issue to fix. Every error and anomaly gets investigated and fixed, you get a PR.

Restore Hyperdrive pool size in checkout-edge

Imagecoreplane/checkout-edge#142PolylaneCritical severityOpened 3 hours agoCI passing
ImageDismissMerge
Files changedInvestigation
src/hyperdrive/pool.ts
@@ -14,21 +14,29 @@ export class HyperdrivePool {
1414 #inFlight = 0;
1515
16- constructor(private readonly binding: Hyperdrive) {
17- this.#ceiling = DEFAULT_CEILING;
16+ constructor(
17+ private readonly binding: Hyperdrive,
18+ private readonly ceiling = poolCeiling(binding),
19+ ) {
20+ if (this.ceiling < MIN_CEILING) {
21+ throw new PoolMisconfigured(this.ceiling, MIN_CEILING);
22+ }
1823 }
1924
2025 async acquire(signal: AbortSignal): Promise<Connection> {
21- while (this.#inFlight >= this.#ceiling) {
22- await this.#queue.wait();
26+ if (this.#inFlight >= this.ceiling) {
27+ queueDepth.record(this.#queue.size, { binding: this.binding.id });
28+ if (this.#queue.size > this.ceiling * QUEUE_ALARM_RATIO) {
29+ logger.warn("hyperdrive pool saturated", {
30+ ceiling: this.ceiling,
31+ waiting: this.#queue.size,
32+ });
33+ }
34+ await this.#queue.wait(signal);
2335 }
2436 this.#inFlight += 1;
2537 return this.binding.connect();
2638 }
@@ -41,9 +53,13 @@ export class HyperdrivePool {
4153 release(connection: Connection): void {
4254 this.#inFlight -= 1;
43- this.#queue.next();
55+ queueDepth.record(this.#queue.size, { binding: this.binding.id });
56+ this.#queue.next();
4457 }
4558
46- get saturated(): boolean {
47- return this.#inFlight >= this.#ceiling;
59+ get saturated(): boolean {
60+ return this.#inFlight >= this.ceiling;
4861 }
4962}
@@ -58,14 +71,18 @@ function poolCeiling(binding: Hyperdrive) {
58- const configured = binding.maxConnections;
59- if (!configured) return DEFAULT_CEILING;
60- return configured;
71+ const configured = binding.maxConnections ?? DEFAULT_CEILING;
72+ // The binding's ceiling is the contract: a worker that opens more
73+ // connections than the pool allows queues here instead of at the origin,
74+ // which is invisible in origin latency. Fail loudly at boot instead.
75+ return Math.max(configured, MIN_CEILING);
6176}
6277
63-export const DEFAULT_CEILING = 5;
64-export const QUEUE_ALARM_RATIO = 4;
78+export const DEFAULT_CEILING = 50;
79+export const MIN_CEILING = 25;
80+export const QUEUE_ALARM_RATIO = 0.5;
6581
66-const queueDepth = noopCounter();
67-const logger = console;
82+const queueDepth = meter.createHistogram("hyperdrive.queue_depth");
6883
Ask about this pull request

It fully maps your app, from cloud to code. All services, repos and providers in one context graph that powers everything.

Works with:

  • Image
  • Image
  • Image
  • Image
  • Image
  • Image
  • Image
  • Image
  • Image
  • Image
  • Image
Topology
Search your cloud resources...
Galaxy
Imagecheckout-edgeWorker
CriticalCritical to your architecture.

Issue hotspot: 4 issues in the last 7 days

Change hotspot: 12 changes in the last 7 days

Cloudflare · coreplane · earth·Compute
Imagehd-prodHyperdrive
CriticalCritical to your architecture.

Change hotspot: 7 changes in the last 7 days

Cloudflare · coreplane · earth·Databases
Imageingest-eventsLambda function
StandardImportant to your architecture.

Issue hotspot: 2 issues in the last 7 days

AWS · coreplane-prod · us-east-1·Compute
Imagepayments-dbDatabase
CriticalCritical to your architecture.

Issue hotspot: 1 issue in the last 7 days

Change hotspot: 3 changes in the last 7 days

PlanetScale · coreplane · us-east·Databases

It prevents slop from hitting prod. All code changes, from bots and engineers, get reviewed against live telemetry.

github.com/coreplane/orders-api/pull/482

ImageAdd trigram index for order search #482

Openrvidal wants to merge 1 commit into main from order-search-trgm
Conversation1Commits1Checks2Files changed1
polylanebotcommented 2 minutes ago···
Caution

Merging this pull request may degrade production (high impact).

Merging this blocks every write to orders while the index builds. migrations/0114_order_search_trgm.sql:3 adds CREATE INDEX … USING gin (search_text gin_trgm_ops) without CONCURRENTLY, and a plain CREATE INDEX takes a full write lock on orders for the whole build. Checkout sustains ~38 writes/s on that table; each one queues behind the lock until the build finishes.

To make this safe: build the index with CREATE INDEX CONCURRENTLY outside the transactional migration.

orders-db · writes per second · last 48h
02040
-48h-24hnow
every one of these writes blocks while the index builds

Polylane analysed de91b47 for production impact.

Some checks were not successful1 failing and 1 successful check
ci / testSuccessful in 2m 4sDetails
Polylane production impactNon-concurrent index build locks writes on ordersDetails
Merging is blocked

It's always available to answer your question. It knows your app and will tirelessly dig through data for you.

Image# engineering
MessagesAdd canvasFiles & links
Image
boris2:09 AM

checkout feels slow. is that us or payments-api?

Image
PolylaneAgent2:09 AM

Us. checkout-edge Wall Time P99 is 18× its own baseline and the requests are queuing for a Hyperdrive connection, not on a downstream call.payments-api is answering in 180ms, flat for a week. Deploy 9f3c2a1 at 2:02 AM shrank the hd-prod pool from 50 connections to 5.

👀 2
Image
aleksandr2:11 AM

how long has it been queueing like that?

Image
PolylaneAgent2:11 AM

Nine minutes, since the deploy landed. It never went above 2 before that. I submitted a PR fix and tagged@justin to deploy 2 minutes ago.

hd-prod · pool queue depth · last 30mdeploy 9f3c2a1
02040
-30m-15mnow
every request in that queue is a checkout waiting

Open in PolylaneConfigure

And it shares insights with other agents. A new production context layer for your coding agents, over MCP or the CLI.

Imageclaude — ~/checkout-edge

> make region a required field on the checkout request schema

●polylane · searchcallers of POST /checkout

●polylane · runToolqueryLogs · request shapes

⏺Hold on: cart-svc and edge-gateway still send region-less requests,41,200 in the last 24h. Requiring it now would 400 them both.

Safer: default it, migrate the two callers, then require.

Are you okay with this plan?

❯ 1. Yes, default it and migrate both callers

❯ 2. Yes, but require it now and let them 400

❯ 3. No, tell Claude what to do differently (esc)

No migration. No new SDKs. Polylane plugs into what you already run.

ImageAWSImageCloudflareImageVercelImageFly.ioImageRenderImageKubernetesImagePlanetScaleImageSupabaseImageModalImageRailwayImageConvexImageClickHouseImageTursoImageDatadogImageSentryImageHoneycombImageAxiomImageGrafana CloudImageBetter StackImageOpenStatusImageLogfireImageGitHubImageSlackImageLinearImageCursorImageDevinImageFactoryImageConductorImageMCP And more →

Built by engineers who carried the pager. From Cloudflare, Webflow, Groq, Twilio, Uber, and Robinhood.

CloudflareCloudflareWebflowWebflowGroqGroqTwilioTwilioUberUberRobinhoodRobinhood
“I spent years building observability platforms, at Baselime and then at Cloudflare. Our tooling is still terrible at finding what's broken, and it can't fix anything on its own. On-call is still broken. I'm fixing it.”
Boris Tane
Boris Tane
Founder

We're making software operate itself. Join us.

Trust is non-negotiable.

AICPA SOC 2

SOC 2 Type II

Independently audited security controls.

ISO 27001

ISO 27001:2022

The international standard for information security management.

End-to-end encryption

AES-256 at rest, TLS 1.2+ in transit.

Data isolation

Fully isolated per organization.

View trust center

Get started with Polylane

Polylane console

Get started for free

Polylane CLI

Common questions.

What is Polylane?

Polylane makes your software self-operating. AI agents learn everything you run, fix what they can, review every code change against the production it deploys to, answer the questions on-call gets, and give your coding agent the same production context.

How does Polylane detect issues?

Agents read your metrics, logs and traces on a cadence and judge them against how each resource normally behaves. Anything your team already charts becomes a check.

Does Polylane change production without review?

No. Every write pauses for your approval with the exact request on screen, and code changes arrive as pull requests that your review and your CI gate.

What does Polylane work with?

AWS, Cloudflare, Vercel, Fly.io, Render, Kubernetes, PlanetScale, Railway, Supabase, Modal, Convex, ClickHouse, and Turso, plus GitHub, Slack, and your observability tools: Datadog, Honeycomb, Axiom, Grafana Cloud, Sentry, Better Stack, OpenStatus, and Logfire.

When can I use Polylane?

Now. One command installs the CLI, connects your coding agents, and creates your account: run curl -fsSL https://polylane.com/setup | bash on macOS or Linux.

Can AI agents use Polylane?

Yes. There is a REST API at api.polylane.com and an MCP server at mcp.polylane.com/mcp, and this site is machine-readable at polylane.com/llms.txt.