# Latest

**URL:** https://community.fly.io/latest.md

[Latest](https://community.fly.io/latest.md) · [Categories](https://community.fly.io/categories.md) · [Tags](https://community.fly.io/tags.md)

---

## [Fly.io Support: community vs. email, 2026 edition (read this first)](https://community.fly.io/t/fly-io-support-community-vs-email-2026-edition-read-this-first/27768)

<div class="topic-metadata">

**Author:** [@injoong](https://community.fly.io/u/injoong)\
**Replies:** 0\
**Last updated:** [May 1, 2026, 9:06pm UTC](https://community.fly.io/t/fly-io-support-community-vs-email-2026-edition-read-this-first/27768 "2026-05-01T21:06:41Z")

</div>

Hello! It’s been a minute and we wanted to give a li’l update about how Fly.io staff will be interacting with the community forum. The big change here is that certain forum posts will now be showing up directly in front…

---

## [Machines network policies: can egress rules cover ICMP and the internal DNS resolver? And is self-test traffic OK?](https://community.fly.io/t/machines-network-policies-can-egress-rules-cover-icmp-and-the-internal-dns-resolver-and-is-self-test-traffic-ok/28776)

<div class="topic-metadata">

**Author:** [@SpaceLightning](https://community.fly.io/u/SpaceLightning)\
**Replies:** 3\
**Last updated:** [October 13, 2026, 9:38pm UTC](https://community.fly.io/t/machines-network-policies-can-egress-rules-cover-icmp-and-the-internal-dns-resolver-and-is-self-test-traffic-ok/28776 "2026-10-13T21:38:13Z")

</div>

Hi all, I run untrusted, short-lived workloads on Machines inside an app with its own private network. The goal is that nothing a workload sends leaves the Machine. Two questions: Network policy scope. As far as I can…

---

## [Sprite restarts by itself 1 to 3 minutes after waking from idle (reproducible, sbd storage)](https://community.fly.io/t/sprite-restarts-by-itself-1-to-3-minutes-after-waking-from-idle-reproducible-sbd-storage/28796)

<div class="topic-metadata">

**Author:** [@Clinton](https://community.fly.io/u/Clinton)\
**Replies:** 1\
**Last updated:** [October 13, 2026, 7:25pm UTC](https://community.fly.io/t/sprite-restarts-by-itself-1-to-3-minutes-after-waking-from-idle-reproducible-sbd-storage/28796 "2026-10-13T19:25:53Z")

</div>

I have a Sprite that restarts by itself shortly after it wakes from idle, and I can reproduce it most times. Posting the details in case staff can see what is bringing it down. All times are UTC. The Sprite is named fore…

---

## [fly-proxy sends TCP keepalive probes every ~5 s on idle client connections, costly for cellular IoT](https://community.fly.io/t/fly-proxy-sends-tcp-keepalive-probes-every-5-s-on-idle-client-connections-costly-for-cellular-iot/28809)

<div class="topic-metadata">

**Author:** [@EricM](https://community.fly.io/u/EricM)\
**Replies:** 2\
**Last updated:** [October 13, 2026, 6:56pm UTC](https://community.fly.io/t/fly-proxy-sends-tcp-keepalive-probes-every-5-s-on-idle-client-connections-costly-for-cellular-iot/28809 "2026-10-13T18:56:25Z")

</div>

We run an MQTT broker on Fly (TCP service on 8883 with handlers = \["tls"\], MQTT keepalive 270 s). Our clients are battery-powered LTE-M devices. The Fly edge sends a zero-length TCP keepalive probe on every idle client …

---

## [Does fly-proxy ever write buffered HTTP request bodies to disk, swap or temp files?](https://community.fly.io/t/does-fly-proxy-ever-write-buffered-http-request-bodies-to-disk-swap-or-temp-files/28794)

<div class="topic-metadata">

**Author:** [@ChrisM74291](https://community.fly.io/u/ChrisM74291)\
**Replies:** 4\
**Last updated:** [October 13, 2026, 3:59pm UTC](https://community.fly.io/t/does-fly-proxy-ever-write-buffered-http-request-bodies-to-disk-swap-or-temp-files/28794 "2026-10-13T15:59:24Z")

</div>

Hello, I’m trying to understand how fly-proxy handles HTTP request bodies before they reach an application, particularly buffering for retries or replays. Could someone from Fly clarify the current behavior? Is the …

---

## [x86 Architecture Guarantee](https://community.fly.io/t/x86-architecture-guarantee/28811)

<div class="topic-metadata">

**Author:** [@mpross34](https://community.fly.io/u/mpross34)\
**Replies:** 4\
**Last updated:** [October 13, 2026, 3:19pm UTC](https://community.fly.io/t/x86-architecture-guarantee/28811 "2026-10-13T15:19:49Z")

</div>

We’re deploying a single image to a machine via “image=” property within our fly.toml file. We know in advance that image is compiled only for x86 CPU architectures; it will not run on ARM architectures. Thus far, no pr…

---

## [Config file not found. Check the file name and the current working directory.](https://community.fly.io/t/config-file-not-found-check-the-file-name-and-the-current-working-directory/28810)

<div class="topic-metadata">

**Author:** [@ShanefromLucidly](https://community.fly.io/u/ShanefromLucidly)\
**Replies:** 0\
**Last updated:** [October 13, 2026, 1:39pm UTC](https://community.fly.io/t/config-file-not-found-check-the-file-name-and-the-current-working-directory/28810 "2026-10-13T13:39:05Z")

</div>

(topic deleted by author)

---

## [Ephemeral secrets deployment](https://community.fly.io/t/ephemeral-secrets-deployment/28739)

<div class="topic-metadata">

**Author:** [@fungs](https://community.fly.io/u/fungs)\
**Replies:** 4\
**Last updated:** [October 13, 2026, 12:03pm UTC](https://community.fly.io/t/ephemeral-secrets-deployment/28739 "2026-10-13T12:03:59Z")

</div>

When deploying secrets from a third-party secret store to fly.io using fly secrets import, old or renamed secrets remain. Is there any recommended and simple way to clear all secrets before deploying new ones? A simple f…

---

## [Sprite Can’t Connect: “Error: failed to start sprite command: Storage not ready”](https://community.fly.io/t/sprite-can-t-connect-error-failed-to-start-sprite-command-storage-not-ready/28785)

<div class="topic-metadata">

**Author:** [@alvis](https://community.fly.io/u/alvis)\
**Replies:** 4\
**Last updated:** [October 13, 2026, 10:58am UTC](https://community.fly.io/t/sprite-can-t-connect-error-failed-to-start-sprite-command-storage-not-ready/28785 "2026-10-13T10:58:20Z")

</div>

My Sprite sprite-15c3de51-2ee8-473c-8580-0acfb4355786 is stuck cold, and both sprite console and sprite exec fail with Error: failed to start sprite command: Storage not ready I saw that Fly previously identified a volu…

---

## [Sprites on 0.0.2-beta.5 reboot repeatedly without a request and recover their journals each boot](https://community.fly.io/t/sprites-on-0-0-2-beta-5-reboot-repeatedly-without-a-request-and-recover-their-journals-each-boot/28806)

<div class="topic-metadata">

**Author:** [@epiphanylab](https://community.fly.io/u/epiphanylab)\
**Replies:** 2\
**Last updated:** [October 13, 2026, 2:24am UTC](https://community.fly.io/t/sprites-on-0-0-2-beta-5-reboot-repeatedly-without-a-request-and-recover-their-journals-each-boot/28806 "2026-10-13T02:24:03Z")

</div>

The Sprites in our org run the same tenant container. The ones on runtime 0.0.2-beta.5 reboot far more often than the one on 0.0.1-rc48, and every boot recovers an ext4 journal. expansion-tenant-t005 (0.0.2-beta.5, cre…

---

## [Machines OpenAPI spec omissions](https://community.fly.io/t/machines-openapi-spec-omissions/27255)

<div class="topic-metadata">

**Author:** [@praneetloke](https://community.fly.io/u/praneetloke)\
**Replies:** 31\
**Last updated:** [October 13, 2026, 1:00am UTC](https://community.fly.io/t/machines-openapi-spec-omissions/27255 "2026-10-13T01:00:13Z")

</div>

Hello! I have found errors in the OpenAPI spec. I am hoping someone from the dev team can address these. The following GET endpoints are missing path parameter definitions under parameters: /apps/{app\_name}/certificate…

---

## [iptables xt\_owner unavailable, secure egress isolation alternatives?](https://community.fly.io/t/iptables-xt-owner-unavailable-secure-egress-isolation-alternatives/28808)

<div class="topic-metadata">

**Author:** [@CMALAVASI](https://community.fly.io/u/CMALAVASI)\
**Replies:** 2\
**Last updated:** [October 12, 2026, 11:32pm UTC](https://community.fly.io/t/iptables-xt-owner-unavailable-secure-egress-isolation-alternatives/28808 "2026-10-12T23:32:32Z")

</div>

Hello Fly.io community, I’m deploying a security-sensitive outbound HTTP fetching service on Fly Machines, using a Go API wrapper and Smokescreen as an HTTP proxy. The architecture requires the wrapper process (UID 100…

---

## [local backup from mpg cluster fails](https://community.fly.io/t/local-backup-from-mpg-cluster-fails/28777)

<div class="topic-metadata">

**Author:** [@bwhitford](https://community.fly.io/u/bwhitford)\
**Replies:** 2\
**Last updated:** [October 12, 2026, 8:41pm UTC](https://community.fly.io/t/local-backup-from-mpg-cluster-fails/28777 "2026-10-12T20:41:39Z")

</div>

I have read the docs and tried the ‘fly proxy / pg\_dump’ process several times with no success. % fly proxy 5433:5432 -a appname % pg\_dump -h localhost -p5433 -U postgres -F c dbname\_db \> dump.sql My pg\_dump command c…

---

## [JNB Capacity issues](https://community.fly.io/t/jnb-capacity-issues/28804)

<div class="topic-metadata">

**Author:** [@rvzyl](https://community.fly.io/u/rvzyl)\
**Replies:** 1\
**Last updated:** [October 12, 2026, 2:44pm UTC](https://community.fly.io/t/jnb-capacity-issues/28804 "2026-10-12T14:44:01Z")

</div>

Hi, we are hitting failed to launch VM: no capacity available in jnb errors. Can you please increase JNB capacity?

---

## [Sprite failed to start: Storage not ready](https://community.fly.io/t/sprite-failed-to-start-storage-not-ready/28798)

<div class="topic-metadata">

**Author:** [@tyanboot](https://community.fly.io/u/tyanboot)\
**Replies:** 2\
**Last updated:** [October 12, 2026, 8:58am UTC](https://community.fly.io/t/sprite-failed-to-start-storage-not-ready/28798 "2026-10-12T08:58:24Z")

</div>

Hi, My Sprite sprite-98720adf-c1d6-4b9b-ae9c-1fb36c4182eb cann’t start, both url and console failed with “Storage not ready”. Does this releated to upgrade from 0.0.1-rc48to0.0.2? Could someone help restore my sprite a…

---

## [Again sprite dead, third time, different app](https://community.fly.io/t/again-sprite-dead-third-time-different-app/28803)

<div class="topic-metadata">

**Author:** [@emileb](https://community.fly.io/u/emileb)\
**Replies:** 0\
**Last updated:** [October 12, 2026, 7:17am UTC](https://community.fly.io/t/again-sprite-dead-third-time-different-app/28803 "2026-10-12T07:17:01Z")

</div>

Can you guys either give real SLA’s on the sprite storage, or can you refund me? https://bouwtel-\*

---

## [Sprites cert issue?](https://community.fly.io/t/sprites-cert-issue/28800)

<div class="topic-metadata">

**Author:** [@dcdc](https://community.fly.io/u/dcdc)\
**Replies:** 1\
**Last updated:** [October 12, 2026, 12:45am UTC](https://community.fly.io/t/sprites-cert-issue/28800 "2026-10-12T00:45:41Z")

</div>

Already checked status page: All \*.sprites.app subdomains are failing the TLS handshake (since at least 2026-10-11 23:40 UTC, still failing at 2026-10-12 00:11 UTC). Every subdomain fails, including made-up ones: 5 sp…

---

## [New machines that mount a Staged secret via files are destroyed within 1 second](https://community.fly.io/t/new-machines-that-mount-a-staged-secret-via-files-are-destroyed-within-1-second/28795)

<div class="topic-metadata">

**Author:** [@hazemsalem](https://community.fly.io/u/hazemsalem)\
**Replies:** 1\
**Last updated:** [October 11, 2026, 1:21am UTC](https://community.fly.io/t/new-machines-that-mount-a-staged-secret-via-files-are-destroyed-within-1-second/28795 "2026-10-11T01:21:11Z")

</div>

In two apps, any new machine that mounts a Staged secret as a file is destroyed by flyd about 1 second after creation. There is no error and no log output. Config changes to the one existing machine are also ignored, and…

---

## [Sprite CLI: Disable upgrade checks](https://community.fly.io/t/sprite-cli-disable-upgrade-checks/28790)

<div class="topic-metadata">

**Author:** [@jonstacks13](https://community.fly.io/u/jonstacks13)\
**Replies:** 2\
**Last updated:** [October 10, 2026, 4:58pm UTC](https://community.fly.io/t/sprite-cli-disable-upgrade-checks/28790 "2026-10-10T16:58:04Z")

</div>

Hi! It looks like since sprites moved more into the fly.io docs, it might not be monitored as much so hoping to link the feature request here: Is it possible to disable the upgrade check? · Issue #216 · superfly/sprites-…

---

## [Unable to authorize access: must authenticate with GitHub](https://community.fly.io/t/unable-to-authorize-access-must-authenticate-with-github/28793)

<div class="topic-metadata">

**Author:** [@Legado](https://community.fly.io/u/Legado)\
**Replies:** 1\
**Last updated:** [October 10, 2026, 3:06pm UTC](https://community.fly.io/t/unable-to-authorize-access-must-authenticate-with-github/28793 "2026-10-10T15:06:17Z")

</div>

Good Morning, I get this message " Unable to authorize access: must authenticate with GitHub account with access to organization 1 (got \[\])" and I’ve seen the different threats talking about it, but yesterday night eve…

---

## [How to retrieve boot-stage diagnostics for a deleted ephemeral Machine?](https://community.fly.io/t/how-to-retrieve-boot-stage-diagnostics-for-a-deleted-ephemeral-machine/28787)

<div class="topic-metadata">

**Author:** [@madeinwill369](https://community.fly.io/u/madeinwill369)\
**Replies:** 0\
**Last updated:** [October 10, 2026, 5:23am UTC](https://community.fly.io/t/how-to-retrieve-boot-stage-diagnostics-for-a-deleted-ephemeral-machine/28787 "2026-10-10T05:23:44Z")

</div>

We have a short-lived Fly Machine used as an isolated BuildKit builder. One admitted attempt created the Machine, then it exited before it reached a started state. We captured an exit event but no boot-stage marker, exit…

---

## [Anthropic Sprite Connector](https://community.fly.io/t/anthropic-sprite-connector/28786)

<div class="topic-metadata">

**Author:** [@darrenl](https://community.fly.io/u/darrenl)\
**Replies:** 7\
**Last updated:** [October 10, 2026, 2:28am UTC](https://community.fly.io/t/anthropic-sprite-connector/28786 "2026-10-10T02:28:12Z")

</div>

Hey all, can someone tell me if the Anthropic connector for Sprites is meant to be used with the Claude CLI agent? I configured the connector, restarted my Sprite, and it keeps saying: Not logged in \* Please run /login …

---

## [An HTTP POST through the Sprite URL doesn't wake a paused Sprite (GET does)](https://community.fly.io/t/an-http-post-through-the-sprite-url-doesnt-wake-a-paused-sprite-get-does/28774)

<div class="topic-metadata">

**Author:** [@mischiefdev](https://community.fly.io/u/mischiefdev)\
**Replies:** 4\
**Last updated:** [October 10, 2026, 12:38am UTC](https://community.fly.io/t/an-http-post-through-the-sprite-url-doesnt-wake-a-paused-sprite-get-does/28774 "2026-10-10T00:38:01Z")

</div>

Summary Main issue. While a Sprite is paused, an HTTP POST to \<sprite-url\>/mcp (https://\<sprite\>-\<orgid\>.sprites.app) gets no response until the client times out. That happened in 6 of 6 tries: 90 s timeout ×4, 45 s ×2,…

---

## [claude-opus-5-5 needs 2.1.280 or newer in sprites](https://community.fly.io/t/claude-opus-5-5-needs-2-1-280-or-newer-in-sprites/28775)

<div class="topic-metadata">

**Author:** [@anuragit15](https://community.fly.io/u/anuragit15)\
**Replies:** 1\
**Last updated:** [October 9, 2026, 6:27pm UTC](https://community.fly.io/t/claude-opus-5-5-needs-2-1-280-or-newer-in-sprites/28775 "2026-10-09T18:27:00Z")

</div>

The model works when called directly. The check is on the Claude Code version, so any agent running Claude Code inside a sprite can’t use the newest Opus. Every other model we use still works on 2.1.251. We lock sprite …

---

## [Blue-green deploy seems to keep adding more machines](https://community.fly.io/t/blue-green-deploy-seems-to-keep-adding-more-machines/28768)

<div class="topic-metadata">

**Author:** [@danmartens](https://community.fly.io/u/danmartens)\
**Replies:** 8\
**Last updated:** [October 9, 2026, 3:10pm UTC](https://community.fly.io/t/blue-green-deploy-seems-to-keep-adding-more-machines/28768 "2026-10-09T15:10:39Z")

</div>

I’ve switched one of our projects to the bluegreen deploy strategy, but now it seems like it it never “cleans up” old machines. I want to have 1 web process that stops after there is no activity for a few minutes (this i…

---

## [Fly Proxy resets a stalled HTTP stream after ~30s on edge and Flycast hops; http\_options.idle\_timeout has no effect](https://community.fly.io/t/fly-proxy-resets-a-stalled-http-stream-after-30s-on-edge-and-flycast-hops-http-options-idle-timeout-has-no-effect/28755)

<div class="topic-metadata">

**Author:** [@nanodeath](https://community.fly.io/u/nanodeath)\
**Replies:** 3\
**Last updated:** [October 9, 2026, 12:40am UTC](https://community.fly.io/t/fly-proxy-resets-a-stalled-http-stream-after-30s-on-edge-and-flycast-hops-http-options-idle-timeout-has-no-effect/28755 "2026-10-09T00:40:27Z")

</div>

Hello fly.io team. I’ve been trying to get Navidrome+Symfonium working smoothly while hosted on fly.io, and it mostly works, but has some unexpected drops when streaming: the connection sometimes drops around 30 seconds …

---

## [Has anyone come up with a good solution for proper scale-to-zero on public sprites? (Dealing with internet scanners)](https://community.fly.io/t/has-anyone-come-up-with-a-good-solution-for-proper-scale-to-zero-on-public-sprites-dealing-with-internet-scanners/28773)

<div class="topic-metadata">

**Author:** [@nihakue](https://community.fly.io/u/nihakue)\
**Replies:** 0\
**Last updated:** [October 8, 2026, 5:11pm UTC](https://community.fly.io/t/has-anyone-come-up-with-a-good-solution-for-proper-scale-to-zero-on-public-sprites-dealing-with-internet-scanners/28773 "2026-10-08T17:11:33Z")

</div>

Does anyone have any good suggestions for dealing with public sprites getting woken by both traffic? My costs are being hugely inflated by bots hitting /login, /wp, /cms, etc. by bots looking for vulnerabilities. I can p…

---

## [Notification when machine state changed](https://community.fly.io/t/notification-when-machine-state-changed/28772)

<div class="topic-metadata">

**Author:** [@flobishop](https://community.fly.io/u/flobishop)\
**Replies:** 1\
**Last updated:** [October 8, 2026, 3:38pm UTC](https://community.fly.io/t/notification-when-machine-state-changed/28772 "2026-10-08T15:38:10Z")

</div>

Is it possible to be notified when machine state changed ?

---

## [App stdout lost for ~20 minutes while the process kept running (lhr), plus a slow machine restart](https://community.fly.io/t/app-stdout-lost-for-20-minutes-while-the-process-kept-running-lhr-plus-a-slow-machine-restart/28770)

<div class="topic-metadata">

**Author:** [@lauren94](https://community.fly.io/u/lauren94)\
**Replies:** 1\
**Last updated:** [October 8, 2026, 3:38pm UTC](https://community.fly.io/t/app-stdout-lost-for-20-minutes-while-the-process-kept-running-lhr-plus-a-slow-machine-restart/28770 "2026-10-08T15:38:00Z")

</div>

Region: lhr · Machine: e82d10e6f3e778 · Size: shared-cpu-1x, 512MB On 7 Oct 2026, every stdout line from this machine between 09:46:22 and 10:06:46 UTC is missing, from both fly logs and the Grafana/Loki log search. Th…

---

## [the \_acme-challenge CNAME had been proxied by Cloudflare, so renewal could never validate](https://community.fly.io/t/the-acme-challenge-cname-had-been-proxied-by-cloudflare-so-renewal-could-never-validate/28712)

<div class="topic-metadata">

**Author:** [@supo](https://community.fly.io/u/supo)\
**Replies:** 4\
**Last updated:** [October 8, 2026, 12:29am UTC](https://community.fly.io/t/the-acme-challenge-cname-had-been-proxied-by-cloudflare-so-renewal-could-never-validate/28712 "2026-10-08T00:29:00Z")

</div>

App kastov, org nicolas-615, hostname supo.live. The apex certificate expired 2026‑09‑25 17:50 UTC (my fault — the \_acme-challenge CNAME had been proxied by Cloudflare, so renewal could never validate). I fixed the DNS …

[Next page](https://community.fly.io/latest.md?page=1)
