Closed Routhinator closed 4 years ago
Oof, didn't realize the default was 1 second. That's a bit too low for Synapse. I've upped it to 5 seconds, but you may want to raise it/lower it further.
Yeah, I have been watching this more and you may also want to add
failureThreshold: 3
I think when the server is in a burst of http traffic from federating it may be throttling incoming connections, which causes the connect timeout. I'll try the update you made first though.
The default failureThreshold
is already 3 (at least on minikube 1.18 where I'm testing right now). If your cloud provider has overriden it, feel free to change it back.
(Looks like it's also the official default in the API spec)
Oh, derp. Sorry. That was right in front of my face on the docs too. :P
Stable without a restart for 12 minutes so far. I think that's got it - thanks.
Now that I have my federation working thanks to the ingress improvements.. My container is crashlooping.. I couldn't figure out why at first but I realized that with all the work it's doing chatting with new federation partners the container is being killed by Kubernetes because it's not responsive enough:
It might need to be given a failure allowance for these circumstances.