Maximilian Hoffman
Maximilian Hoffman
The update on this is, I think this is a rare GC-stats interaction, where GC removes a chunk that it's deemed unreachable before stats tries to read it. We made...
@robertabcd This panic has no impact on data correctness. I am still working on trying to plug the hole, hopefully will have a fix soon.
@robertabcd This PR appears to stabilize the script https://github.com/dolthub/dolt/pull/9106. I've been running it for awhile and the throughput is much higher, and I have not seen panics yet. Working the...
It's been running smoothly for a couple hours now, though I'm still seeing a spike in latency that I can't explain ``` flush done in 44.35 seconds flush done in...
The fix PR is almost ready, waiting on a colleague on vacation to sign off when he's back Wednesday. I was running this over the weekend and some of the...
I'm seeing one other notable thing related to GC. Every `dolt_gc` adds a new tablefile to our oldgen store. Each tablefile in oldgen is indexed/accessed separately from the others, and...
This is pretty safe to merge, I do think the way you plan to use the AST might give you some issues. Normally we would fold execution time decision making...
> Sometimes it's nice to be able to access the parsed query, say in a `Table`'s callbacks; for example, if one implements a table backed by a system that allows...
@wk8 Closing this issue for now pending more details. We are happy to reopen, just let us know more about what you are trying to do.
#benchmark