Fluxer - Scheduled maintenance for real-time infrastructure improvements – Maintenance details

All systems operational

Show current status
JUN14

Scheduled maintenance for real-time infrastructure improvements

CompletedJune 14, 2026 at 1:30 AM – 3:58 AMUTC

Affects

Gateway

Under maintenance from 1:30 AM to 3:58 AM

  • CompletedJune 14, 2026 at 3:58 AMUTC

    Maintenance has completed successfully. It took some time to identify an edge case where if you had a DM open with the system user ("Fluxer"), you were actually not reconnected properly, hence the presence oscillation. We're now all back again. Thanks for your patience!

  • UpdateJune 14, 2026 at 3:36 AMUTC

    All users have reconnected at this point, but we're diagnosing a bug with rapid presence oscillation. Member lists have been temporarily disabled in a select few guilds.

  • UpdateJune 14, 2026 at 3:00 AMUTC

    We're monitoring recovery.

  • UpdateJune 14, 2026 at 2:19 AMUTC

    We identified a bug with our new hotpatching system that was not caught in testing and are fixing this before finalising the rollout.

  • UpdateJune 14, 2026 at 1:50 AMUTC

    We're restoring access to the app as quickly as capacity allows. Hang in there!

  • In progressJune 14, 2026 at 1:30 AMUTC
    Maintenance is now in progress.
  • PlannedJune 11, 2026 at 4:44 AMUTC

    The gateway has been running smoothly for the past 10 days, with no major issues.

    We've now got scheduled maintenance planned to move our gateway infrastructure onto a new code version. For context, that infrastructure is a cluster of 54 nodes spread across six specialised tiers for redundancy and load distribution, and it's served us very well. The new version introduces a system for safe, durable, and strongly consistent code updates that can roll out across the whole cluster in seconds, all through a standardised process that minimises human error. It also stabilises the wider deployment pipeline, should we ever need to carry out a rolling deployment of our stateful tiers.

    Our lead scientist, Rick Sanchez, reckons it'll be a quick twenty-minute adventure, in and out, though you might want to take that with a pinch of salt. The work we've put into boosting our backend's performance and helping it absorb load means we can promise a faster recovery time than on previous occasions.

    Once this new durable hotpatch deployment system is in place, we shouldn't ever need to "restart Fluxer" again, unless something catastrophic triggers cascading failures in our real-time stack. We're aiming for that 99.99% uptime!