Charith Ellawala
Charith Ellawala
Fixed in #1959. Thanks @psolarcz
Nice idea!
@rcrowe I am assuming that the important thing here for you is that all Cerbos instances have the same set of policies to provide consistent answers in the event of...
Yes, that sums it up quite nicely. I think we can achieve both use cases with some sort of "strategy" configuration that lets you define when/how to fallback to the...
How about just having a single `failoverThreshold` setting? ```yaml storage: driver: overlay overlay: baseDriver: blob overlayDriver: sqlite3 failoverThreshold: 3 ``` The value of `failoverThreshold` is the number of errors seen...
In `union` mode, shouldn't both stores be healthy and available because they are treated as a single store? When a request comes through, I'd first try to find it in...
How can we deal with the case where a policy exists in one store and its dependencies exist in the other store? So far, I have been thinking that we...
Yeah, even with the policies, I am inclined to treat each store as self-contained because loading dependencies from a different store feels a bit iffy. I wonder whether we should...
Let's start with the driver itself being immutable regardless of what the underlying stores are. If there's a compelling reason to support writes, we can revisit it. Our current mutable...
Fixed in #1960 and #1968