This repository has been archived by the owner on Apr 4, 2023. It is now read-only.
Move Cassandra pilot creation to ScaleOut #322
Comments
Closed
wallrj
added a commit
to wallrj/navigator
that referenced
this issue
Apr 26, 2018
For each nodepool StatefulSet: * Create a number of pilots to match the number of pods that will be created for the StatefulSet. * Delete higher index pilots which have been left behind after the statefulset has been scaled in. * Do not delete a pilot if there is a pod with a matching name, (that pod won't be able to decommission its self unless it can read its desired configuration from its pilot) * Do not delete a pilot unless it is owned by the cluster that is being synchronised. (this is not an expected state, but we don't want to delete anything unless it was created by us) Fixes: jetstack#322
wallrj
added a commit
to wallrj/navigator
that referenced
this issue
Apr 27, 2018
For each nodepool StatefulSet: * Create a number of pilots to match the number of pods that will be created for the StatefulSet. * Delete higher index pilots which have been left behind after the statefulset has been scaled in. * Do not delete a pilot if there is a pod with a matching name, (that pod won't be able to decommission its self unless it can read its desired configuration from its pilot) * Do not delete a pilot unless it is owned by the cluster that is being synchronised. (this is not an expected state, but we don't want to delete anything unless it was created by us) Fixes: jetstack#322
wallrj
added a commit
to wallrj/navigator
that referenced
this issue
May 9, 2018
For each nodepool StatefulSet: * Create a number of pilots to match the number of pods that will be created for the StatefulSet. * Delete higher index pilots which have been left behind after the statefulset has been scaled in. * Do not delete a pilot if there is a pod with a matching name, (that pod won't be able to decommission its self unless it can read its desired configuration from its pilot) * Do not delete a pilot unless it is owned by the cluster that is being synchronised. (this is not an expected state, but we don't want to delete anything unless it was created by us) Fixes: jetstack#322
|
We're trying a generic pilot controller instead: #353 |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
I think it make sense to create pilots during scaleout rather than as a separate control.
And then in #285 we can remove the pilot during scale down.
The text was updated successfully, but these errors were encountered: