Looking at the code of k8s api call to get namespaces, seems we are reading in batches of 500 by default. Meaning we’ll read first 500 items and check a “continue” flag in response.
If it is true, then continue reading the next 500, until done. This is not only for namespaces, but also for listing deployments/pods etc.
Seems the same code works OK in OpenShift 4.x, but not OpenShift 3.11, so maybe OpenShift 3.11 doesn’t return the “continue” flag properly. We tried to increase the batch size as a workaround, and verified we did get more than 500 items back.
Asking for an enhancement to better support this
Release : SAAS
Component : Integration with APM
A fix was made to the UMA stable release to better handle batch reading. This supports UMA 3.x and 4.x.