Please use this template for troubleshooting questions.
My issue:
Seeing substantial latency apparently added by nginx-ingress. Upstream returns data (can be sizable, 20MB) in double digit ms, response time to client is 30sec+. This is new behavior in the last few days, as always no one did anything.
Tests from inside cluster to application via ingress show multisecond averages, tests to application via service are consistently sub-second.
Default controller config, using master/minion with no master and minimal minion annotations (nginx.org/client-max-body-size: 10m).
Shared ingress with multiple (~50) services served, only this one complaining.
I do see in logs responses get buffered but my understanding is this is due to slowness in delivering to client but given differences between ingress and direct service I have some doubts. No other error messsages of note with default (not debug) logging, someone has a minion missing a master but that’s just noise.
Not sure what to look at/try next.
Version of NIC and/or NGINX:
5.1.4
Deployment environment:
AKS