Is LaunchDarkly Down? How to Tell and What to Do

LaunchDarkly is the feature flag management layer for teams running continuous delivery. When LaunchDarkly goes down, your SDKs fall back to cached or default flag values — which means features may silently activate or deactivate for all users without any intentional release.

Statusfield brings reported LaunchDarkly incidents into the same workspace as your other monitors. Select supported components and route relevant changes to your team. Vendor reports can lag an incident, so check your own application as well.

Follow LaunchDarkly incidents with your team

Statusfield brings reported status for LaunchDarkly together with your other critical tools. Choose the changes that should alert your team. Eligible new accounts can start a free trial — no credit card required.

Start free trial

Try any paid plan for 30 days, no credit card required. Free trial for first-time customers.

Common Signs of a LaunchDarkly Outage

Symptom	What's likely happening
SDKs failing to initialize or connect	LaunchDarkly streaming infrastructure is degraded
Flag evaluations returning default values for all users	SDK fell back to defaults; streaming connection lost
Feature flags not updating after changes in the dashboard	Streaming delivery is degraded
LaunchDarkly dashboard not loading	Web infrastructure is affected
REST API calls returning 5xx errors	API layer is unavailable
Experiments and A/B tests paused	Experimentation infrastructure is affected
Relay Proxy failing to connect to LaunchDarkly	Backend connectivity to LD infrastructure is degraded
Audit log and flag change history not updating	Event pipeline is stalled

How to Confirm It's LaunchDarkly, Not Your SDK

1. Check LaunchDarkly's streaming endpoint

LaunchDarkly SDKs connect to a streaming endpoint to receive real-time flag updates. A degradation here means flags won't update even if the dashboard shows changes.

bash

curl -I "https://stream.launchdarkly.com/all" \
  -H "Authorization: YOUR_SDK_KEY"

A connection timeout or 5xx confirms the streaming infrastructure is degraded.

2. Test the REST API

bash

curl "https://app.launchdarkly.com/api/v2/flags/default" \
  -H "Authorization: YOUR_ACCESS_TOKEN"

A 503 or 429 (unusual rate limiting) confirms API-level issues.

3. Check SDK initialization logs

Most LaunchDarkly SDKs log initialization status. Look for:

"Streaming connection failed"
"Using cached flag values"
"SDK failed to connect"

If you see these during normal operation, your SDK is operating in degraded mode.

4. Try a direct flag evaluation

javascript

const flagValue = ldClient.variation("my-flag", userContext, false);

If you're getting the fallback false for all users when your dashboard shows the flag enabled, the SDK is not receiving current flag state from LaunchDarkly.

The Impact of LaunchDarkly Outages on Your Product

LaunchDarkly outages are particularly tricky to diagnose because the system is designed to degrade gracefully — SDKs fall back to cached values or default values. This means:

Features may silently change state: If the cache is empty (first deployment) and LaunchDarkly is down, all flags return defaults — which may differ from your intended rollout state.
Experiments stop collecting data: A/B test flag evaluations during an outage produce unreliable experiment data.
Staged rollouts freeze: A gradual rollout that was progressing will stop advancing until connectivity recovers.
Kill switches don't work: If you rely on a flag to disable a broken feature, that kill switch won't fire reliably during a LaunchDarkly outage.

Follow reported LaunchDarkly status in Statusfield and choose which changes reach your configured alert destinations. Delivery depends on source updates, polling and successful processing. Check the catalog for supported components and the pricing page for channel availability.

Follow LaunchDarkly incidents with your team

Follow reported incidents for LaunchDarkly in Statusfield and send relevant updates to your configured channels. Eligible new accounts can start a free trial — no credit card required.

Start free trial

Try any paid plan for 30 days, no credit card required. Free trial for first-time customers.

What to Do During a LaunchDarkly Outage

Immediate (first 5 minutes)

Determine what state your flags are in — check your SDK logs to understand whether it's using cached values or defaults. This tells you what behavior your users are experiencing.
Identify any kill switches or high-risk flags — if you have a kill switch for a new feature that's in production, confirm it's in the right state in cache. If not, prepare to deploy a code-level disable.
Pause any active staged rollouts — LaunchDarkly won't advance percentage rollouts while disconnected from the backend.
Alert your release team — any planned deploys that depend on flag changes should be held until connectivity recovers.

Communication (first 15 minutes)

Notify your engineering team of the LaunchDarkly outage so no one spends time debugging flag behavior as a code bug.
If user-facing behavior has changed due to fallback flag values, assess whether it requires a customer communication.
Hold any feature releases that depend on LaunchDarkly for the rollout mechanism.

Recovery

After LaunchDarkly recovers, verify that all SDKs reconnected and are receiving current flag state (check SDK logs for "streaming connection re-established").
Review any experiments running during the outage — the data from the degraded window is likely unreliable.
Confirm staged rollouts are at the expected percentage and have not reset.

LaunchDarkly Components and What Each Outage Looks Like

Component	Impact when degraded
Streaming delivery	Flags don't update in real time; SDKs use cached values
REST API	Flag reads/writes from the API fail
Web dashboard	UI inaccessible; can't change flags
Experimentation	A/B test data not collected; experiment metrics unreliable
Event pipeline	Flag evaluations not logged; analytics incomplete
Relay Proxy	On-premise proxy can't fetch updated flag state
Mobile delivery	Mobile SDK flags use stale cache
Edge SDKs	Edge flag evaluation degraded

Why DIY Flag Monitoring Misses Degraded Delivery

Checking whether your application is running isn't enough to know if LaunchDarkly is healthy:

Graceful fallback hides the problem: A well-configured SDK looks healthy even when disconnected — it just uses cached or default values silently.
No error in your application logs: Unless you've explicitly instrumented SDK connection state, your application logs won't reflect a LaunchDarkly connectivity issue.
The dashboard may work while delivery is degraded: The LaunchDarkly web UI can be available while the streaming delivery infrastructure for SDKs is down — you can change flags but they won't propagate.

Effective LaunchDarkly monitoring checks the streaming endpoint health, not just application availability.

Bring LaunchDarkly updates to your team

Eligible first-time customers can start a free trial, no credit card required.

Follow LaunchDarkly incidents with your team

Follow LaunchDarkly and your other vendors in one Statusfield workspace. Select supported components and route the updates your team needs. Eligible new accounts can start a free trial — no credit card required.

Start free trial

Try any paid plan for 30 days, no credit card required. Free trial for first-time customers.

FAQ

Is LaunchDarkly down right now?

This article is a troubleshooting guide, not a live outage verdict. Check the Statusfield catalog and the official vendor status page for the latest reported status, affected components and observation times. Compare those reports with what your own application is experiencing.

How can I get alerts about LaunchDarkly incidents?

Does an operational LaunchDarkly status mean my application is working?

No. A vendor status page describes reported service conditions. Your account, integration, network or application can still fail. Check your own public HTTPS endpoint and use logs or workflow tests to investigate the customer experience.

Browse the service catalog and plans and alert destinations.

Is LaunchDarkly Down? How to Tell and What to Do