Add support for Pod Disruption Budget

[josephteddick]

This PR intends to allow users to make OpenCost highly available by configuring a Pod Disruption Budget (affinity and replicas already present).

Fix for issue: #286

[mittal-ishaan]

Thank you @josephteddick

[ameijer]

@josephteddick , does this mean when you run multiple replicas, you have multiple prom scrape targets, one for each OC? PR LGTM, just trying to understand

[josephteddick]

@josephteddick , does this mean when you run multiple replicas, you have multiple prom scrape targets, one for each OC? PR LGTM, just trying to understand

Nope, each replica has the same setup in that they are pointing to the service name for Prometheus. It's then up to internal routing for which pods (replica) it will go to and fetch the metrics (assuming Prometheus is also HA). You also don't have to worry about duplication because each replica will have it's own PVs.

High Availability here is mostly for the OpenCost API in that we can always hit it to get allocation metrics. Hope that helps.