← Back to the carnage
DIGITALOCEAN
🔴 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)
// The official story
Major Service Interruption in MKC1
// Status page says
Aug 19 , 12:22 UTC Identified - We have identified the root cause of the issue in the MKC1 region affecting multiple racks and nodes. Our Engineering team is actively implementing remediation steps to restore connectivity and bring the impacted nodes back online. During this time, customers may continue to experience disruptions to GPU workloads and Serverless Inference. Kubernetes (DOKS) worker nodes may also remain in a NotRea
🔗 Official status page (for the PR-approved version)
https://status.digitalocean.com/incidents/qd8v4wddyqjrDigitalocean's Greatest Hits
🟢 DIGITALOCEAN
Oct 03, 2026
⚪ BLR1 region went full Sleeping Beauty from 11:59-13:02 UTC, ghosting Monitoring, App Platform, Databases, Functions, Load Balancers, Spaces, and the control panel. Users got the classic combo of latency, dropped connections, and errors—basically a networking horror movie where the power came back before anyone got to the climax. (Auto-resolved: no updates in 24h)
🟢 DIGITALOCEAN
Sep 16, 2026
🟡 DigitalOcean's support portal decided to ghost users because some third-party vendor flipped the wrong switch. Can't submit tickets, can't get help—basically the digital equivalent of calling 911 and reaching a busy signal. Support team is probably on Slack wondering why their own tools don't work. (Auto-resolved: no updates in 24h)
🟢 DIGITALOCEAN
Sep 09, 2026
🟡 DigitalOcean's Atlanta datacenter found out what happens when you skip the electrician's calls—GPU performance taking a sad nap while redundant power limps along. Your ML models are now running at 'laptop charger plugged into a gas station outlet' speeds. (Auto-resolved: no updates in 24h)
🟢 DIGITALOCEAN
Jul 23, 2026
🟡 DigitalOcean's Serverless Inference is serving nothing but HTTP 500s from the llama model. Turns out even AI gets the zoomies and decides not to show up for work. Engineers are frantically whispering to the model about deadlines. (Auto-resolved: no updates in 24h)