> cat /var/log/failures.log
A comprehensive record of when things went wrong.
🟢 REDIS
Feb 07, 2026
🟡 Redis Azure customers discovering that 'cloud' apparently means 'sometimes invisible.' Network gremlins are loose and engineers are pretending to know where they came from. (Auto-resolved: no updates in 24h)
🟢 MONGODB
Feb 07, 2026
🟡 MongoDB's Azure West US deployment decided nodes were overrated. Customers watching their clusters report for duty—or not. Cluster modifications now moving at geological speeds. (Auto-resolved: no updates in 24h)
🟢 JIRA
Feb 07, 2026
⚪ JIRA and JSM ghosted users for a bit on Feb 6, then suddenly reappeared claiming everything was fine. Atlassian's version of 'my phone died, sorry' - but for an entire ticketing system. (Auto-resolved: no updates in 24h)
🟢 JIRA
Feb 05, 2026
⚪ JIRA and JSM ghosted us-east-1 users for 29 minutes before remembering where they work. 🍃 Atlassian's way of saying 'time zone drift' = infrastructure amnesia. (Auto-resolved: no updates in 24h)
🟢 SUPABASE
Feb 05, 2026
🔴 Someone shipped a landmine in [email protected] and Edge Functions walked right into it. The fix? Pretend v2.95.0 never existed and rollback to v2.94.1 like it's a bad relationship. If you manually pinned the cursed version, congrats, you're the QA team now. (Auto-resolved: no updates in 24h)
🟢 FLY
Feb 05, 2026
🟠 MPG clusters in Amsterdam, DC and Singapore are having an identity crisis thanks to a downstream Machines API meltdown. Most users just got slower replicas; unlucky few can't reach their primary at all. Fly's firefighters are on scene. (Auto-resolved: no updates in 24h)
🟢 FLY
Feb 05, 2026
🟠 Turns out 'degraded' is Fly's polite way of saying their database replicas forgot how to do their job. MPG clusters in Amsterdam and DC are having an identity crisis—some can't even find their primary node. The Machines API decided to pull the fire alarm, and everyone's pretending this was intentional. (Auto-resolved: no updates in 24h)
🟢 DATADOG
Feb 05, 2026
🟠 Datadog's monitor evaluations decided to process at geological timescales. They've identified the problem and are 'implementing a fix'—which in status-page speak means the oncall engineer finally stopped screaming into the void. (Auto-resolved: no updates in 24h)
🟢 GRAFANA
Feb 05, 2026
🟡 Grafana's Hosted Metrics in us-central-0 decided to play dead for queries and rule evaluations. Stale data is making a comeback tour while engineers dust off their detective hats. (Auto-resolved: no updates in 24h)
🟢 SUPABASE
Feb 05, 2026
🟡 Some US customers can't resolve Supabase domains, but their actual databases are running fine. It's like having a working car with a broken GPS—the engine's great, you just can't find it. Only specific DNS providers affected, because consistency is overrated. (Auto-resolved: no updates in 24h)
🟢 SUPABASE
Feb 04, 2026
🟡 Supabase's instance operations went full Sleeping Beauty—new projects won't spin up, paused ones refuse to budge. The good news: your existing projects are still breathing. Bad news: they're watching the chaos unfold. (Auto-resolved: no updates in 24h)
🟢 ANTHROPIC
Feb 04, 2026
⚪ Claude briefly forgot how to be Claude between 16:20-16:55 UTC. Engineers found the issue, applied the digital equivalent of smelling salts, and everything's back to pretending it never happened. (Auto-resolved: no updates in 24h)
🟢 GITHUB
Feb 03, 2026
⚪ GitHub's CI/CD runners decided to take an unscheduled sabbatical, leaving queued jobs to timeout staring at an empty queue. The incident lasted nearly 6 hours across standard and larger runners. Your deployment dreams were deferred, but at least they're back now. (Auto-resolved: no updates in 24h)
🟢 SENDGRID
Feb 03, 2026
🟠 SendGrid's Marketing Campaigns and API had a brief existential crisis, returning 500 errors like a teenager who forgot their homework was due. Single Sends, Automations, and contact uploads all decided to ghost users simultaneously. The good news: actual email delivery kept working, so at least the mail got through even if everything else was on fire. (Auto-resolved: no updates in 24h)
🟢 CONFLUENCE
Feb 03, 2026
🟡 Confluence and Jira Mobile's OAuth theater just went dark—turns out 'identity service' means identity crisis. Forge apps are also having an existential moment while engineers frantically search their memory banks for why they trusted this. (Auto-resolved: no updates in 24h)
🟢 JIRA
Feb 03, 2026
🟡 Atlassian's identity service decided to stop recognizing people, breaking OAuth for Jira and Confluence mobile users plus Forge apps. Engineers are investigating—which corporate-speak means they're still figuring out which log file to check first. (Auto-resolved: no updates in 24h)
🟢 MUX
Feb 01, 2026
🟡 Mux's own monitoring system decided to play possum. Ironic considering they help you watch videos—turns out they can't watch their own systems in real-time. Investigating why the watchers can't be watched. (Auto-resolved: no updates in 24h)
🟢 AMPLITUDE
Jan 31, 2026
{
"skip": true
}
```
This is scheduled maintenance that completed successfully. Per the rules, we skip planned/scheduled maintenance notifications. (Auto-resolved: no updates in 24h)
🟢 BUILDKITE
Jan 31, 2026
🟡 Someone else's database caught fire and took Test Engine with it. Buildkite's playing the blame game while engineers refresh status pages obsessively. (Auto-resolved: no updates in 24h)
🟢 PLANETSCALE
Jan 31, 2026
🟡 PlanetScale's Query Insights decided visibility was overrated and ghosted everyone. Your database queries are still running, but they're flying blind now—like piloting with the dashboard ripped out. (Auto-resolved: no updates in 24h)