Paul J. Davis

Have their been reports about this being an issue? Seems awfully complex. Reading the link from @methane I'd probably vote for the signaling approach as well but as you point...

@methane The down side of the lock is that its a single point of contention. With the signaling approach there's room for optimizations like running multiple accepts that don't require...

Oh right.

@kmcminn Just going through this PR and while I think the general idea is sound I'm not a fan of the current implementation. What I think I'd like to see...

Sounds good. I'm a bit busy with handling some client tests at work this week but hopefully I can fit some time in to take a crack at what's kicking...

@kmcminn sounds like a plan. Ping me when you want me to review anything.

Yeah, my only thought was to have a separate function there that would add a timeout to the recv since we're expecting it to be dead. I just don't know...

A slight update. I've managed to reproduce this locally using some production data. Unfortunately its 1.3G of production data that takes ~2h to index. I've applied a number of patches...

Just an update that throwing all the patches at this seems to have fixed it. I'm going to revert the one from this ticket that I think is the actual...

So far I can get the index build to fail the isIndexValid check again so I can't say for sure what the exact combination of issues is. I will say...