Pod object and Pod’s health

Couple of questions here: a) what the concept of Pod object satisfies? There is the pause container which holds the namespace for the actual ‘app’ container(s) that will get added later and use that namespace. So, what operations in real life qualify to have more than one containers running in a Pod ? Can you provide to the point examples? What are considered sidecar operations after all?

b) what is the threshold for e.g. N containers in the Pod that master will be able to tell that Pod is not healthy? So basically, how many containers in a Pod need to be down to mark the Pod down?

the most common example would be some sort of logging sidecar. Depending on your implementation or your service you may want a service that manipulates the files for the pod etc. Some databases can use sidecars for log shipping in replication, or you might see some process to clean up log files as well. Although available, those options are not suggested specifically because there are more alternatives.

StackExchange DevOps Q#4248, answer score: 1