Erik Grinaker

I'd probably automate it so that it could be run e.g. nightly, but haven't given it much thought really.

I believe this is caused by the P2P address handling being fundamentally broken, and e.g. conflating the bind (listen) address with the advertised address and the external address. I am...

Afraid I'm not familiar with the details here either, @brapse and @milosevic would know more. But the verification bit happens here, I guess: https://github.com/tendermint/tendermint/blob/ee993ba8ff9893ab4c46b269987a3138ead223ff/blockchain/v2/processor.go#L135 As I understand, it currently just...

Yeah, great find @arulajmani, thanks for tracking this down! > in #87167 did we actually hit the "extremely rare OOM hazard"? We do seem to get a very large stack...

>I'd like to iterate on this in a followup to ensure that the range keys are eventually GC'd and dropped from the LSM. We may want to backport a simple...

>Sorry, I've been working on personal mac for open source work for better ergonomics, and I keep pushing stale branches. Pushed the updated branch. I see, thanks. Looked it over,...

>Maybe this test is too aggressive with the import cancellations. After a while, the imports slowed to a crawl. There's barely any CPU utilization, but it's all in CheckSSTConflicts. I'd...

Right. I notice that we don't use range key masking in `CheckSSTConflicts`. I think we could, because the range tombstone would necessarily have to be above any point keys (so...

Btw, it would be good to assert MVCC stats here too. We could do something like the clearrange roachtest: https://github.com/cockroachdb/cockroach/blob/aac830251139eca515ae31053414bb7ab624a115/pkg/cmd/roachtest/tests/clearrange.go#L82 But since this only runs the check on splits/merges and...

Could be #73376, which keeps popping up. Unfortunately we may not get around to addressing it for 23.1, but we're considering bumping the priority.