Tobias Grieger

Thanks for the input, I'll consider that for my January 2025 sprint!

@palmskog thanks, that sounds like the problem. Seeing that the code isn't going to be used in production any time soon (ever?) I agree that it's probably not worth fixing...

> I can't speak for the original Verdi team, but production-ready code is generally not a goal in academic projects, nor encouraged to any significant degree by stakeholders (funding agencies,...

Hi Bryan, was just poking around the source code for exactly this, glad you've already given it a look. Would you recommend I give the above a try using your...

Thanks @tidwall and apologies for not getting back to you earlier. I didn't realize you were sending reads through the Raft log in `high`, but that makes sense. > When...

> In both cases "dropped" and "apply multiple times" should be covered because high requires a new log index for all read/write commands prior to responding to the client. I'm...

Ah, I see. One caveat here is that when you receive an error, the proposal could still have applied. That is, the burden of figuring out what happened is shifted...

Consider the following: - Node 1 is leader in Term 11 and successfully sends appends to the other two nodes (but does not hear back) - Node 2 campaigns and...

From what I'm hearing in CC, the 1 minute timeout will not reliably drain the node. But no timeout will; there seem to be cases where we don't drain successfully....

Have the same issue, but wanted to add that even if the text showed up, why is the extra click needed? The list of notifications disappears every time I click...