Jamie Brandon

> that they might just be negative lookups on the transfer path. They happen during compaction, not during prefetch/commit, so they're definitely compaction misses. During prefetch it's almost all cache...

https://github.com/tigerbeetledb/tigerbeetle/issues/560 does seem to indicate that compactions are trashing the grid cache. > We can insert compaction blocks at low-priority and let CLOCK figure out what's worth keeping, according to...

I reran with a 16gb grid cache (and verified the increased memory usage). Here's the last big compaction.  3s long. ~6k misses vs ~4k hits. The database file is...

Some napkin math: https://hur.st/bloomfilter/?n=1e9&p=0.001&m=&k= suggests that if we wanted to keep grid reads down to a few per batch, we'd need 1-2gb of bloom filter for 1e9 transfers. Presumably breaking...

With a 16gb cache this seems much better:  (Unfortunately this doesn't improve performance more than a few % because throughput is totally dominated by compaction) I'd like to know...

The first idea that comes to mind is to change the api: * In the caller, store `?BlockPtr` instead of `BlockPtr`. * On `grid.write_block` set the block to null and...

Producing graphs in a low-dependency fashion is tricky. Probably the best option is r + base, which is fairly stable over time.

I don't think fossils tools can be used on git repos. There are a bunch of code review tools that work on top of github. I haven't tried any of...

Extra wishlist: 7. Update conversations in real-time so I don't constantly have comment skew (github only updates comments when the page is refreshed).

reviewable.io hits 2 and 3. Sort of 4 (has an expand-all in easy reach) and sort of 5 (by virtue of not having a conversation view, instead there's a next/prev...