John Spray
John Spray
> Start with the PurgeGarbage subcommand as it doesn't use stream_tenants. If you could implement a streaming listing in GenericRemoteStorage to enable this, it would also be useful in places...
@skyzh can you take a look at this one when you're back in the office?
Let's look over our existing backpressure-related issues and make a plan - https://github.com/neondatabase/neon/issues/7317 - https://github.com/neondatabase/neon/issues/5897 - https://github.com/neondatabase/neon/issues/5415
Plan: - Prioritize: https://github.com/neondatabase/neon/issues/7317 - For the l0 stack problem: this may depend on other compaction design decisions? Potentially trigger compaction on LSN rather than time -- that way faster...
@ololobus did you mean to pause this? The test is still unstable
The PR says: "Please notice that most likely it will not fix flukyness of test_subscriber_restart.", so I'd be pretty confident that this isn't the fix.
Right, I was trying to fast-forward past the "observe it for a few days" part.
I played with moving the `enum{}` wrapper up into Timeline, and it works, but the bulk of the code churn remains of course in changing all the infallible code paths...
Heikki added a "no response from pageserver for N seconds" thing recently, maybe we can hook into that for a metric?
Related issue that would benefit from similar observability: https://github.com/neondatabase/cloud/issues/23515