esatterwhite
Kubernetes: Added a helm chart to bootstrap skyring to a kubernetes c…
Changed PR To add option for Mongo for persistency. Also I do have a question for @esatterwhite . In this chart I do not seed using 3456. Is there a reason in the examples why you seed on both 3455 and 3456?
there is nothing specific about the port numbers themselves. They were just easy to type and didn't conflict when running a cluster locally.
I generally like to start a cluster with more than one seed node to make sure a stable communication ring is established before allowing more nodes to join. It isn't necessarily required though.
also, if you are running on separate hosts or in containers / k8s. the ports numbers specifically are a little less important - every node could just use <HOST>:3456 and be fine
@esatterwhite I'm trying to add support for scylladb in this chart as well due to the design being more resilient to node failures. However when I'm trying to use the scylla configs the skyring nodes don't actually connect to one another. The environment variables I am using are
DEBUG: '*'
channel__host: 'node-1'
nats__hosts: 'nats-a:4222,nats-b:4222,nats-c:4222'
seeds: 'node-1:3455'
storage__backend: '@skyring/scylladown'
storage__path: 'skyring-1'
storage__collectionPoints: 'scylla-1:9042,scylla-2:9042,scylla-3:9042' (this line will be replaced with the headless service for scylla in Kubernetes
Is there something obviously wrong I am doing here? Scylladb is supported oob here right? The nodes appear to be hanging on the read stream for the recovery.