> cat /var/log/failures.log
A comprehensive record of when things went wrong.
🟢 AMPLITUDE
Aug 25, 2026
🟡 New users staring at a blank login page since 16:45 PT—turns out the gateway to analytics was just a void. Existing sessions? Still alive and well, thanks for asking. The irony of a data company losing its own onboarding data isn't lost on us. (Auto-resolved: no updates in 24h)
🟢 WISE
Aug 24, 2026
🟡 Wise is currently investigating why their international money transfer service decided to take an unscheduled siesta. Customers attempting to move funds are experiencing delays—because what's a fintech without a little financial thriller? Engineers are probably stress-testing their coffee machine as we speak. (Auto-resolved: no updates in 24h)
🟢 CONTENTFUL
Aug 22, 2026
{
"skip": true
}
```
This is scheduled maintenance, which we're skipping per the rules. (Auto-resolved: no updates in 24h)
🟢 GCP
Aug 20, 2026
⚪ us-west1 region decided to take an unscheduled nap from 8:40 AM to 12:20 PM, treating timeouts like they were going out of style. Engineers emerged from the war room with the all-clear, though 'mitigation complete' is corporate for 'we turned it off and back on.'. (Auto-resolved: no updates in 24h)
🟢 MAILGUN
Aug 20, 2026
🟠 Google's West Coast datacenter decided to have an existential crisis, and Mailgun's juggling incoming mail like a one-armed circus performer. Log processing is moving at continental drift speeds while some messages are just... thinking about their life choices. (Auto-resolved: no updates in 24h)
🟢 LANGSMITH
Aug 20, 2026
🟡 LangSmith's build pipeline decided new revisions aren't worth processing. Control plane is having an existential crisis in GCP US. Engineers are investigating why their deployment factory forgot how to manufacture deployments. (Auto-resolved: no updates in 24h)
🟢 VERCEL
Aug 20, 2026
🟡 Vercel hit a snag where the Dashboard can't query runtime logs—classic case of selective amnesia. Build logs and ingestion still working, so at least they didn't burn the whole house down. Engineers are currently arguing with their own infrastructure. (Auto-resolved: no updates in 24h)
🟢 FLY
Aug 20, 2026
🟠 Fly's ord-0 Kubernetes cluster decided to take an unscheduled nap, and now MPG clusters are stuck in an existential crisis trying to restart. The team's frantically Googling 'why won't my cluster wake up' while your deployments contemplate the void. (Auto-resolved: no updates in 24h)
🟢 BOX
Aug 20, 2026
⚪ Box's entire nervous system misfired for 5 minutes—login, files, uploads, downloads, API all playing dead simultaneously. The good news: it's over. The bad news: it happened at all. Engineers are pretending this was totally intentional. (Auto-resolved: no updates in 24h)
🟢 BOX
Aug 20, 2026
⚪ Box's file browser decided to take an unscheduled nap from 3:44-3:49pm PDT, serving users errors and molasses. The patient has recovered, though some folks might still be experiencing phantom pains. If your files are still playing hide-and-seek, Box support awaits your distress call. (Auto-resolved: no updates in 24h)
🟢 TAILSCALE
Aug 20, 2026
🟡 Tailscale's coordination service is having an identity crisis—admin console and core coordination are both partially down. Engineers are pretending to understand the problem while the service contemplates why it even exists. (Auto-resolved: no updates in 24h)
🟢 BOX
Aug 20, 2026
🟡 Box users trying to open Office docs are getting the digital equivalent of a busy signal. The integration is having what we in the industry call 'commitment issues.' Team is investigating why their integration suddenly went nonverbal. (Auto-resolved: no updates in 24h)
🟢 OPENAI
Aug 20, 2026
🔴 OpenAI's crown jewel went full Fort Knox—signups and logins dead in the water. Users refreshing their browser like it's a vending machine that ate their money. Engineers frantically searching for the 'undo' button in production. (Auto-resolved: no updates in 24h)
🟢 MIXPANEL
Aug 19, 2026
🟠 Mixpanel's event properties decided to play hide-and-seek in their dropdowns and nobody told them the game's over. Engineers are currently having a very productive staring contest with their monitors while your data takes an unscheduled vacation. (Auto-resolved: no updates in 24h)
🟢 CURSOR
Aug 19, 2026
🟡 Cursor's Fable 5 model went full Rip Van Winkle, leaving engineers scrambling to explain why their fancy AI forgot how to AI. Time to dust off those backup models while the team figures out what went wrong. (Auto-resolved: no updates in 24h)
🟢 DIGITALOCEAN
Aug 19, 2026
🔴 DigitalOcean's MKC1 decided to take an unscheduled sabbatical, taking GPU workloads, Serverless Inference, and Kubernetes nodes along for the ride. Engineering is frantically flipping switches and probably refreshing their monitoring dashboards like they're slot machines. (Auto-resolved: no updates in 24h)
🟢 GOOGLE_WORKSPACE
Aug 19, 2026
⚪ Update: RESOLVED: **Summary**
Some customers may experience slowness while using Google Drive.
**Description
🟢 ZAPIER
Aug 19, 2026
🟡 Zapier's Functions feature decided to ghost users trying to create or edit anything. The good news: regular Zaps still work, so at least half your automation didn't have an identity crisis today. (Auto-resolved: no updates in 24h)
🟢 GOOGLE_WORKSPACE
Aug 18, 2026
⚪ Google's Classroom decided to play hide-and-seek with admins for 3.5 hours. Turns out the new homepage was just being mysterious. All access restored, teachers can resume yelling at students online.
🟢 SENTRY
Aug 18, 2026
🟠 Sentry's GoCD pipelines went full Rip Van Winkle while engineers performed emergency root cause archaeology. The mitigations are supposedly in progress, which in incident speak means they found the smoking gun but haven't figured out how to put it back together yet. (Auto-resolved: no updates in 24h)