We're using Apache Ignite in Kubernetes as a caching layer right now and
it's working well for us so far.
One thing that's been problematic is that when defining a readinessProbe,
the TcpDiscoveryKubernetesIpFinder only sees pods already in a ready state.
This means that if you start all your pods at once (as part of deployment
creation) you'll wind up with each pod being in its own grid.
It seems like others are experiencing this problem and I think I have a
The basic premise is to kill pods after they connect to a grid that doesn't
contain the alphabetical first ip in the service list.
I was wondering if there is a better solution to this segmentation problem
or if this seems workable.