Mender on k8s is very slow to process larger artifacts

We are running a mender instance in a kubernetes cluster using the official helm chart at for version 3.4.

The setup works and we can use the instance, but when we upload an artifact of about 1.5GiB through either the ui or the commandline client the processing of the artifact takes a very long time (easily 20 minutes), while uploading the artifact is only around a single minute. Looking at kubectl top none of the mender services are using large amounts of cpu or ram during the processing. I’ve also tried uploading the artifact to our minio instance directly, which also only took under a minute.

Timeouts and size limits on our ingress are set up properly to allow for uploading a file of this size. The upload also doesn’t fail, it just takes a very long time compared to a local demo setup.

Does anybody here have an idea where this could be coming from, or what we could do to get more debugging info?