Timeout problem when uploading larger files with Traefik

[smoky19x]

Hello,

when uploading larger files there is a timeout after 60 seconds in Traefik with the current community-compose.yaml. The following entry in the “proxy: > command:” section of the file prevents the timeout:

--entryPoints.web.transport.respondingTimeouts.readTimeout=0

Could you please include this entry.

Thank you and best regards.

[gionn]

Setting readTimeout=0 effectively disables the timeout, allowing clients to hold connections indefinitely. This will increases susceptibility to DDOS attacks (e.g. slowloris) and can hang server resources until a forced process restart.

Please set a reasonable upper limit for your use case, the default 60s should be suitable for most scenarios, set it to 600s for specific cases that truly require extended durations.

Not sure we really want to change this setting since our compose files are for reference and not implementing a specific use case.

[smoky19x]

I think many people will run Traefik behind a reverse proxy, so “readTimeout=0” would only affect the connection to the proxy. In the nginx setup before Traefik was introduced, there was no 60 second timeout either. I suspect that some more people might run into the timeout problem on an upload if the value stays at 60 seconds. If you don't want to set the value to 0, it might still make sense to increase it to a reasonable value and add a comment or information that this value may need to be adjusted.

[smoky19x]

It took a while for me to find out why the upload of larger files no longer works, whereas it still worked with nginx as a proxy.

[alxgomz]

@smoky19x for now we have added a task to better document that change (OPSEXP-3014) and help people save time in the future.
Out of curiosity, how big was the upload which failed and how fast was the connection between the client and the server?

Also please keep in mind these docker compose files - while we intend to give them care and attention as much as we can - are not meant to be used for production workload, at least not as-is.

[smoky19x]

We experienced the upload problem with files from approx. 150 MB. The upload bar in Share continued to run, but at the end there was always the error message “Error: 502 Bad Gateway”.

I also realize that you do not declare the docker compose files as “production ready”, but I suspect that many people use these files as a template for their setup. If you search for terms like “alfresco upload problem”, “alfresco upload 502” or similar, you will find a whole lot of entries 😉.