We are currently trying out SelfHosted API gateway on a Kubernetes cluster. The same traffic seem to run ok on the ManagedGateway. However, on the Selfhosted one, the pods quickly increasing in memory usage.
We set our memory limit to 24GB. Each pod started with 100Mb and quickly got to 9GB in an hour under traffic. At which point it suddenly jump to 12GB, then 17GB before crashout with out of memory in the log. CPU usage are only 3% to 4% during the whole time.
I feel like it might be some dumb thing in our configuration but can't seem to find any documentation. Almost like something getting stored in memory and not getting flush out.
We don't configure any persistent storage. We also turn off the stdout logging as well. No caching API. No Appinsight logging.
Any help is much appreciated!!!