This PR aims to solve some problems that we were having with deploying KubeMQ via the existing chart. There are numerous improvements that I'll attempt to outline below:
Add support for nodeSelector, affinity, tolerations: This was required in order for us to place our pods in a deterministic way.
Move token into a secret, add support for providing license data: This is the prescribed k8s way to manage sensitive data. Also provides a facility for offline license validation which was documented, but difficult to find and not supported by the existing chart.
Added support for generating a configmap with arbitrary environment variables. This is useful in defining options that are not directly supported by the helm chart.
Enabled the use of more specific (fullname) resource names. This is pretty common in other charts, and better enables users to define distinct KubeMQ clusters within a common namespace. The default behaviour should be consistent with the existing naming convention by setting both nameOverride and fullnameOverride to kubemq-cluster. Setting these values to "" will use the calculated names.
This PR aims to solve some problems that we were having with deploying KubeMQ via the existing chart. There are numerous improvements that I'll attempt to outline below:
nodeSelector
,affinity
,tolerations
: This was required in order for us to place our pods in a deterministic way.fullname
) resource names. This is pretty common in other charts, and better enables users to define distinct KubeMQ clusters within a common namespace. The default behaviour should be consistent with the existing naming convention by setting bothnameOverride
andfullnameOverride
tokubemq-cluster
. Setting these values to""
will use the calculated names.statefulset
pods.0.2.0
.