---
title: "Fly.io Changelog"
description: "New features, improvements and fixes across Fly.io."
url: "https://fly.io/changelog/"
---

# Fly.io Changelog

- [Rotate MPG user passwords from flyctl](https://fly.io/changelog/rotate-postgres-passwords-flyctl.md) (2026-10-07): Rotate Managed Postgres user passwords from flyctl with fly mpg users rotate-password.
- ['List' endpoints in Machines API now support pagination](https://fly.io/changelog/machines-api-pagination.md) (2026-10-01): The Machines API's list endpoints for Machines, volumes and apps now support pagination.
- [`fly auth logout` is now less likely to bite you](https://fly.io/changelog/fly-auth-logout-fix.md) (2026-09-22): fly auth logout now only revokes your locally stored token, not FLY_API_TOKEN or FLY_ACCESS_TOKEN.
- [Managed Postgres is now in the Machines API](https://fly.io/changelog/managed-postgres-machines-api.md) (2026-09-15): Create, fork, restore and delete Managed Postgres clusters through the Machines API.
- [Cutting Corrosion’s Broadcast Bandwidth by 10x](https://fly.io/changelog/corrosion-bandwidth-reduction.md) (2026-09-02): Corrosion, our state-replication system, now uses about a tenth of the bandwidth, thanks to compression and Plumtree.
- [Egress IP support in Machines API](https://fly.io/changelog/machines-api-egress-ips.md) (2026-09-01): Create and manage app-scoped egress IPs through the Machines API.
- ["Where is my billing page?"; alternative title: making billing page easier to find](https://fly.io/changelog/find-billing-page.md) (2026-08-13): Billing is now linked from settings and the account menu, where people kept looking for it.
- [Tweaks and improvements for PaaS deployments in UI and CLI!](https://fly.io/changelog/bluegreen-deploy-improvements.md) (2026-08-11): Dashboard scale operations show up in activity, and bluegreen deploys in flyctl catch more misconfigurations.
- [You can now restore your MPGv1 backups as a v2 cluster](https://fly.io/changelog/managed-postgres-v1-backup-to-v2.md) (2026-08-11): Restore a Managed Postgres v1 backup into a new v2 cluster from the dashboard.
- [fly mpg restore now supports naming and point-in-time restore](https://fly.io/changelog/managed-postgres-point-in-time-restore.md) (2026-08-10): fly mpg restore can now restore to a point in time and name the restored cluster.
- [Small tweaks to repo-reattachment flow](https://fly.io/changelog/github-repo-reconnect.md) (2026-08-03): Reconnecting a GitHub repo to an app no longer triggers a new deploy, and finds fly.toml outside the repo root.
- [Fly Machines API connector for Sprites](https://fly.io/changelog/sprites-machines-api-connector.md) (2026-08-03): Connect Sprites to the Fly Machines API without putting a Fly token inside the Sprite.
- [Improvements to PaaS UI settings](https://fly.io/changelog/dashboard-app-settings.md) (2026-08-03): More per-app settings are editable in the dashboard, and changes open a PR against your fly.toml if you've connected GitHub.
- [Secrets batch import on UI!](https://fly.io/changelog/dashboard-secrets-bulk-import.md) (2026-07-30): Paste a whole .env file into the dashboard to import secrets in one go.
- [MPGv2: New Regions and GA](https://fly.io/changelog/managed-postgres-v2-ga.md) (2026-07-20): Managed Postgres v2 is in seven more regions and heading to GA. New v1 clusters will be turned off.
- [Machine details page now shows cordon and lease status](https://fly.io/changelog/machine-cordon-lease-status.md) (2026-07-16): The Machine details page now shows whether a Machine is cordoned or leased.
- [MPGv2 has forking in beta](https://fly.io/changelog/managed-postgres-v2-forking.md) (2026-07-16): Fork a Managed Postgres v2 cluster from the dashboard, much faster than restoring a backup.
- [Sprites now run Ubuntu 26.04 LTS](https://fly.io/changelog/sprites-ubuntu-26-04.md) (2026-07-08): New Sprites now run Ubuntu 26.04 LTS.
- [A new home page for your MPG v2 clusters](https://fly.io/changelog/managed-postgres-v2-cluster-overview.md) (2026-07-03): A rebuilt overview page for Managed Postgres v2 clusters.
- [MPGv2 is available in more regions](https://fly.io/changelog/managed-postgres-v2-more-regions.md) (2026-06-30): Managed Postgres v2 is now available in ams, iad, fra and ord.
- [New docs: keeping long-running tasks alive and running private apps with Flycast](https://fly.io/changelog/long-running-tasks-flycast-guides.md) (2026-06-22): Two new guides: keeping background work alive with autostop, and running private apps with Flycast.
- [Managed Postgres V2 beta now in yyz](https://fly.io/changelog/managed-postgres-v2-toronto.md) (2026-06-09): The Managed Postgres v2 beta is now available in yyz as well as dfw.
- [Fly Machines Security Update: DirtyFrag, Fragnesia, and CopyFail](https://fly.io/changelog/kernel-update-dirtyfrag-fragnesia.md) (2026-06-04): Machines now run kernel 6.12.91, which fixes the DirtyFrag and Fragnesia privilege-escalation bugs.
- [HTTP Service editor via dashboard](https://fly.io/changelog/dashboard-http-service-editor.md) (2026-05-26): Edit your app's HTTP service settings, like auto stop/start and minimum running Machines, from the dashboard.
- [Clipboards, mobile machines view, managing scale, minor tweaks!](https://fly.io/changelog/dashboard-copy-buttons-mobile.md) (2026-05-20): Copy buttons all over the dashboard, a mobile-friendly Machines list, and other small fixes.
- [Managed Postgres v2 is now in beta!](https://fly.io/changelog/managed-postgres-v2-beta.md) (2026-05-19): Managed Postgres v2, rebuilt on Fly Machines for better stability, is now in beta.
- [(Unfortunately) We're NAT'ing Fly Machines' IPv6 Addresses](https://fly.io/changelog/machines-ipv6-nat.md) (2026-05-19): Individual Machines no longer get public IPv6 addresses. They keep IPv6 connectivity, and app IPv6 addresses are unchanged.
- [Mini produce: landing page now shows button to latest visited org](https://fly.io/changelog/last-visited-org-shortcut.md) (2026-05-13): The fly.io home page now has a button back to the org you visited last.
- [Scale machines UI improvements and fixes](https://fly.io/changelog/dashboard-machine-scaling.md) (2026-05-08): Scaling Machines from the dashboard now creates a release, and the fly.toml page tells you when a newer version exists.
- [Background restart via dashboard](https://fly.io/changelog/dashboard-background-app-restarts.md) (2026-05-06): App restarts from the dashboard now run in the background, and everyone viewing the app sees them in real time.
- [New guide: Integrating flyctl into automation and CI/CD pipelines](https://fly.io/changelog/flyctl-ci-cd-guide.md) (2026-05-05): A new guide to running flyctl from CI: picking token types, GitHub Actions workflows, monorepos and parsing JSON output.
- [Your posts* are now getting sent directly to our desks](https://fly.io/changelog/community-forum-support-tickets.md) (2026-05-01): Posts in the forum's Questions / Help category now land directly in the support team's ticketing system.
- [fly-replay is now scoped to private networks (by default)](https://fly.io/changelog/fly-replay-private-network-scope.md) (2026-05-01): fly-replay now only routes between apps on the same private network unless your org allows cross-network replays.
- [Copy Fail](https://fly.io/changelog/copy-fail-vulnerability.md) (2026-04-30): Machines and Sprites aren't affected by the Copy Fail vulnerability unless you share one instance between untrusted users.
- [Migrating from Machine-scoped to App-scoped Egress IPs](https://fly.io/changelog/app-scoped-egress-ip-migration.md) (2026-04-21): Promote a Machine-scoped egress IP to an app-scoped one yourself with fly m egress-ip promote, in flyctl v0.4.37 and later.
- [fly-proxy now gracefully terminates WebSocket connections before stopping/suspending a machine](https://fly.io/changelog/websocket-graceful-shutdown-autostop.md) (2026-04-09): fly-proxy now sends WebSocket clients a close frame before a Machine auto-stops or suspends, so they know to reconnect.
- [New guide: Managing Machines with the Machines API](https://fly.io/changelog/machines-api-guide.md) (2026-04-08): A new guide to creating, configuring and updating Machines from your own code with the Machines API.
- [Introducing t3.tigrisfiles.io for public content](https://fly.io/changelog/tigris-public-content-domain.md) (2026-04-02): New Tigris accounts serve public buckets and objects from t3.tigrisfiles.io. Existing URLs keep working.
- [Return to sender: Fallbacks for fly-replay](https://fly.io/changelog/fly-replay-timeout-fallback.md) (2026-03-20): fly-replay has two new fields: a timeout, and a fallback for when the replay target is unhealthy or unavailable.
- [Infra Log Season Two](https://fly.io/changelog/infra-log-relaunch.md) (2026-03-11): The Infra Log is back, with write-ups of what went wrong in our incidents and why.
- [Sprites: Cold Storage Allowance](https://fly.io/changelog/sprites-free-cold-storage.md) (2026-03-09): Every org now gets 20GB of Sprites cold storage free each month.
- [Fly-ifying our internal DNS](https://fly.io/changelog/internal-dns-reliability.md) (2026-03-06): Our per-host DNS resolvers now talk to each other through fly-proxy, so lookups get health checks, retries and fallback routing.
- [Sprites Now Speak MCP](https://fly.io/changelog/sprites-mcp-support.md) (2026-03-04): Sprites now support MCP natively, so your agent can create Sprites, run commands, manage services and take snapshots itself.
- [:authority pseudo-header is now properly forwarded to h2_backend apps](https://fly.io/changelog/h2-backend-authority-header.md) (2026-02-27): Apps using h2_backend now receive the original :authority pseudo-header from fly-proxy.
- [Better load balancing when your nearest region is unhealthy](https://fly.io/changelog/load-balancing-unhealthy-regions.md) (2026-02-26): Fly Proxy now routes around your nearest region faster when it's unhealthy, instead of trying it first every time.
- [Sprites: Plan upgrades/downgrades now possible](https://fly.io/changelog/sprites-plan-upgrades-downgrades.md) (2026-02-25): Change your Sprites plan from the Plan & Billing page. Upgrades apply immediately, downgrades at the end of the month.
- [Easier certs behind Cloudflare, BYO certs, and other goodies](https://fly.io/changelog/custom-certificates-cloudflare.md) (2026-02-17): Verify domains behind Cloudflare with a TXT record, import your own certificates, and manage certs through the Machines API.
- [Proxy Fairness, Take 2 (Or: Why Did I Have Trouble Connecting to Fly in Europe)](https://fly.io/changelog/fly-proxy-fairness-fix.md) (2026-01-28): A fix in fly-proxy so that one app's flood of small WebSocket messages can't slow connections for everyone else in a region.
- [Fixing an Intermittent WebSocket Issue](https://fly.io/changelog/websocket-connection-fix.md) (2026-01-23): We found and fixed a proxy bug that made WebSocket connections sometimes fail to open, mostly hitting Phoenix LiveView apps.
- [HTTP/1.1 trailers support in fly-proxy](https://fly.io/changelog/http-1-1-trailers-fly-proxy.md) (2026-01-21): fly-proxy now passes HTTP trailers through to HTTP/1.1 apps, not just apps using h2_backend.
- [Cheaper Sprites Storage](https://fly.io/changelog/sprites-storage-price-cut.md) (2026-01-20): Sprites storage now costs $0.02/GB-month cold and $0.50/GB-month hot, which makes inactive Sprites much cheaper to keep around.
- [Taking Wireguard out of the Hot Path](https://fly.io/changelog/fly-proxy-tls-instead-of-wireguard.md) (2026-01-19): fly-proxy now forwards traffic between servers over its own mutually authenticated TLS connections instead of WireGuard, removing a bandwidth bottleneck.
- [More Sprites Plans!](https://fly.io/changelog/new-sprites-plans.md) (2026-01-16): New Sprites plans with higher limits on concurrent active Sprites, for teams running more of them at once.
- [Launch your app from the dashboard! Better UI! Heck yeah!](https://fly.io/changelog/dashboard-app-launch-redesign.md) (2026-01-15): Launching an app from GitHub in the dashboard now takes fewer steps and has cleaner config.
- [Vanta support (finally!)](https://fly.io/changelog/vanta-integration.md) (2026-01-08): Orgs with the Compliance package can now connect Vanta, from Vanta or from the Compliance tab in the dashboard.
- [Attaching Managed Postgres Clusters to Apps](https://fly.io/changelog/managed-postgres-app-attachments.md) (2026-01-08): Managed Postgres clusters now show which apps they're attached to, so you can tell what depends on a cluster before you delete it.