Fly.io change history

Fly.io reported a major outage of the fly-metrics.net hosted Grafana and log search instance starting 2026-10-16 18:08 UTC, resolved at 2026-10-16 19:24 UTC (1 h 15 min). The vendor attributed unavailability to host maintenance; logs and metrics shipped to other sources via Fly Telemetry or the Fly Log shipper remained available.

Fly.io’s major fleet-wide host reboots mitigated PwnEscape (Linux Kernel CVE-2026-99999), making Machines, volumes and MPG instances unavailable for 15–20 minutes per host. The incident ran from 2026-10-16 16:22 UTC to 2026-10-17 05:15 UTC (12 h 52 min); it was resolved, though a small number of hosts remained degraded or offline.

Fly.io reported a critical outage affecting Sprite creates in all regions outside Europe, starting 2026-10-08 21:23 UTC and resolved at 2026-10-08 21:54 UTC (31 min). Fly.io applied a fix for a potential issue; Sprite pools recovered and creates resumed succeeding.

Fly.io reported a major outage in LHR (London, UK): regional anycast and static egress IPs were nonfunctional, with anycast traffic rerouted to nearby regions. The incident started at 2026-10-08 05:40 UTC and was resolved at 2026-10-08 06:00 UTC, lasting 19 min. Fly.io identified BGP announcement issues and restored BGP before resolution.

Fly.io reported a critical partial outage in AMS starting 2026-10-03 08:01 UTC, lasting 56 min and resolved at 2026-10-03 08:57 UTC. Offline hosts made some machines unavailable and blocked Managed Postgres v1 and v2 control-plane operations; some MPGv1 clusters later experienced possible degradation. Hosts and all impacted services recovered.

Fly.io reported a critical network outage affecting connectivity to the AMS region, starting at 2026-09-26 19:31 UTC and resolved at 2026-09-26 19:53 UTC (21 min). Fly.io worked with upstream providers to restore connectivity and reported the network recovering at 19:46 UTC.

Fly.io reported a major GRU outage starting 2026-09-24 16:31 UTC, lasting 2 h 40 min and resolved at 2026-09-24 19:11 UTC. Apps continued running, but machine creation or updates and connections could fail. Fly.io linked the issues to a distributed state database migration, identified a bug, and implemented a fix.

Fly.io reported a major incident affecting Private Networking (6PN), starting at 2026-09-24 14:51 UTC and resolved at 2026-09-24 15:37 UTC, lasting 46 min. A fix was implemented by 2026-09-24 15:01 UTC, with private networking error rates returning to normal.

Fly.io reported a major partial Sprites outage with intermittent API failures near SJC, later extending to other regions and affecting creating, connecting to, or managing Sprites. It ran from 2026-09-23 18:44 UTC to 2026-09-24 02:12 UTC (7 h 28 min) and was resolved after mitigations and potential fixes were deployed.

Fly.io reported a major incident involving elevated private networking errors between machines in some regions, starting at 2026-09-23 14:47 UTC and lasting 35 min. A fix was implemented by 2026-09-23 14:56 UTC, and the incident was resolved at 2026-09-23 15:23 UTC.

A major Upstash outage started 2026-09-23 12:42 UTC and resolved 2026-09-24 07:22 UTC (18 h 39 min). Redis connection timeouts affected databases hosted in fra or gig and clients routed through either region, including databases elsewhere. Replicas failed to finish draining during a planned rolling upgrade and remained out of service; a fix was implemented.

Fly.io reported a major IPv6 connectivity incident affecting a subset of hosts in DFW, starting 2026-09-23 06:18 UTC and resolved at 2026-09-23 12:15 UTC, lasting 5 h 57 min. Machines on affected hosts could experience inbound/outbound IPv6 issues; IPv4 was unaffected. Fly.io identified regional maintenance by an upstream internet transit provider.

Fly.io reported a major Sprites API partial outage affecting a subset of Sprites users, starting 2026-09-02 23:03 UTC and resolved at 2026-09-03 00:41 UTC, lasting 1 h 37 min. Error rates decreased during monitoring; no cause or fix was stated.

Fly.io reported a major API background job queue failure starting 2026-09-02 07:21 UTC: creating apps, assigning IP addresses, and creating/renewing certificates could fail. A fix was implemented by 07:31 UTC; the incident was resolved at 07:44 UTC, lasting 22 min.

Fly.io reported a major network incident in the Los Angeles (LAX) region, where apps could experience higher latency or be unreachable. It started at 2026-08-23 01:28 UTC and was resolved at 2026-08-23 02:10 UTC, lasting 41 min. Fly.io stated that upstream networking issues had resolved.

Fly.io reported a major partial outage of MPG (v1) in ORD from 2026-08-20 07:25 UTC to 2026-08-20 07:57 UTC, lasting 32 min. An issue with the ord-0 Fly Kubernetes cluster left many MPG clusters unable to restart; a fix was implemented at 07:32 UTC, and the incident was resolved.

Fly.io’s major secrets service outage blocked new app creation and secret updates; existing apps continued running. It lasted 1 h 3 min, from 2026-08-14 19:30 UTC to 2026-08-14 20:33 UTC. Fly.io cited a failed deployment and failed over the secrets database to a replica, after which the Machines API appeared healthy.

Fly.io reported a major outage in GRU: new MPG clusters could fail to create due to insufficient capacity. It started at 2026-08-04 19:03 UTC and was resolved at 2026-08-05 03:22 UTC, lasting 8 h 19 min; Fly.io tweaked hosts to allow more machine allocation.

Fly.io reported a critical outage with high numbers of 5XX errors on the Machines API and dashboard, starting 2026-07-20 07:10 UTC and resolved at 2026-07-20 17:09 UTC after 9 h 59 min. A fix was implemented; some Managed Postgres v1 clusters were also degraded, while existing machines and Managed Postgres v2 were unaffected.