[Debug help] Function respond 200 but client receive 500

vnourdin commented 1 year ago

Due diligence

I have a long-running function that take 50s and then respond 200, but the client receives a 500 Can't reach service for: my-function.namespace.

My actions before raising this issue

Check the troubleshooting chapter of SfEE
tune timeouts of gateway to 180s
tune exec_timeout of my function to 60s

Expected Behaviour

The function's answer should reach the client without being modified or invalidated by faasd. If this is a timeout problem, the function should be interrupted instead of finishing its job normally.

Current Behaviour

the function finish as expected 08:41:57 GET / - 200 OK - ContentLength: 8B (49.0651s)
the client receives a 500 Can't reach service for: my-function.namespace.
There is no error log on the function logs
in faasd-provider logs I have the following error 08:41:57 error with proxy request to: http://10.62.3.247:8080, Get "http://10.62.3.247:8080": EOF

Are you a GitHub Sponsor (Yes/No?)

[ ] Yes
[x] No

List All Possible Solutions and Workarounds

No idea of workaround nor solution, I'm looking for help to investigate and find out what's happening.

Steps to Reproduce (for bugs)

I have other long-running functions working well, son I don't know how to reproduce.

Your Environment

OS and architecture: Debian bullseye 5.10.149-2 x86_64
containerd: v1.6.8
faasd: 0.16.7

alexellis commented 1 year ago

Hi, it sounds like you are running into a timeout issue.

You can find out how to configure timeouts in the eBook, make sure you have the latest copy from Gumroad.

Alex

alexellis commented 1 year ago

Since I know this is working and have documented it, we'll get this closed. So if you need additional help or support please join the community call: https://docs.openfaas.com/community/

vnourdin commented 1 year ago

So the error with proxy request to ... EOF error is timeout related ? I already had errors like exec-timeout reached, I thought it would be that clear for every timeout ! I'll play a bit with my timeouts config.

By the way, I couldn't find a clear description of each timeout, I have a good idea of how works the exec one but for read and write it's more unclear to me. Is there somewhere an explanation I missed about those ?

openfaas / faasd