Traefik 2.10 repeats requests to the backend multiple times before responding. (No retry middleware or rules configured) #9872
Labels
Milestone
Comments
|
Hello @waffshappen, Thanks for reporting this! We opened PR #9849 to fix this. |
rtribotte
added
kind/bug/confirmed
|
Closed by #9874. |
Can confirm it works, thanks for the fast fix! |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Welcome!
What did you do?
Configured traefik as normal (podman, static .yaml config without access to the podman socket) with a
ntfycontainer as service.I do not have the retry middleware or any configured retry rules on the other middlewares.
Opened https://ntfy.my-domain-here
Sent a Test Message. (PUT against traefik)
What did you see instead?
When running the latest 2.10.0 Traefik container (podman, static yaml configuration) every request forwarded to the final service is sent roughly 10 times before traefik responds. Despite each request responding with a "200".
Problems with that:
Both the browser and access logs show a single request from the client and only traefik then doing the above.
Example jaeger trace (different service, same issue):
What version of Traefik are you using?
Affected: 2.10.0 ( https://hub.docker.com/layers/library/traefik/v2.10.0/images/sha256-369a4f37fd35962e43a93babc7eda11a57bd74fc710bb3aeea0c3b785c867b9a?context=explore )
Not affected: 2.9.X (currently on https://hub.docker.com/layers/library/traefik/v2.9.10/images/sha256-6dee6938b5ebfc511a82ef4e09c80268835a2f67393e5c06c3e4ef9a14d1817b?context=explore )
This happens with every container i tested - nextcloud, forgejo, ntfy and more. And every config, network=host, slirp4netns, pasta
What is your environment & configuration?
Platform
Traefik Configuration
Some values have been replaced with [...] or an $identifier.
All replaced limits are firmly beyond the request sizes.
traefik.yml
routers/tls.yml
routers/middlewares.yml
routers/blockbots.yml
routers/ntfy.yml
catchall fallback domain (For users without SNI)
traefik run command (podman generate systemd managed)
(Godebug is set so that traefik prefers chacha)
Example ntfy container (using https://passt.top/ but happens in slipr4netns containers too)
/usr/bin/podman run \ --cidfile=%t/%n.ctr-id \ --cgroups=no-conmon \ --rm \ --sdnotify=conmon \ -d \ --tmpfs /tmp \ --replace \ --label io.containers.autoupdate=registry \ --hostname ntfy \ --name ntfy \ -p 127.0.0.1:$hostport:80 \ --network pasta \ --tz Europe/Berlin \ -v ntfy_settings:/etc/ntfy \ -v ntfy_data:/var/cache/ntfy \ --blkio-weight=50 \ --memory=2G \ docker.io/binwiederhier/ntfy:latest serveIf applicable, please paste the log output in DEBUG level
Log file with DEBUG level. Other domain logs excluded for brevity.
Excluded lines of other domains (or config dumps) marked with [...].
I read each before removing them, they all appear irrelevant and contain no obvious errors.
The text was updated successfully, but these errors were encountered: