Slower huddle joins in EU region
degradedA media node failed health checks and drained traffic slowly. Joins took up to four seconds in the EU. We rebalanced the pool and shortened the drain window.
System status
Four subsystems. One binary. All green.
Chat, tasks, calendar and meetings run as four subsystems of one deployment. If one is degraded, you'll see it here before you feel it — and dedicated instances report against their own infrastructure.
All systems operational
updated 17:41 · 41,822 events today
Subsystem health
Chat & messagingoperational
p50 46msp95 112ms99.99% uptime
Tasks & boardsoperational
p50 50msp95 112ms99.98% uptime
Calendaroperational
p50 42msp95 112ms99.99% uptime
Meetings & mediaoperational
p50 60msp95 123ms99.96% uptime
Push & activityoperational
p50 38msp95 112ms99.97% uptime
Live event stream
This is the socket you'd be on.
17:41:02chatmessage.delivered # product → 12 clients31ms
17:41:02presencepresence.update sara → online12ms
17:41:03taskstask.moved "Ship voice notes" → done28ms
presence 214 onlinesockets 1,148regions 3degraded 0
Incident history
Ninety days, written up honestly.
Every incident gets a public note with what broke, what we did, and what changed afterwards. Subscribe and they arrive as push, not as a status-page email at 3am.
Push notifications delayed
degradedA queue backlog delayed web push by up to twelve minutes. In-app delivery was unaffected because it does not use the same path. Queue capacity increased and alerting now fires at half the previous threshold.
File uploads rejected for some types
partialA magic-byte signature update wrongly rejected some valid documents. Fixed forward and signatures now ship behind a staged rollout.
Board drag lag on very large boards
minorBoards over roughly 400 cards re-rendered too eagerly while dragging. Not an outage, but it felt like one — virtualised the column render.
Dedicated instances are monitored separately against your own infrastructure — your deployment channel is the fastest place to ask.
Four subsystems, one uptime number.
That's the other half of "one app": one thing to monitor instead of four vendors blaming each other.