vdbergh
vdbergh
> so, that probably means we will have to apply the fix in two steps if that's possible. It is possible I guess. PR1: this PR+comment out locking statement in...
It seems to be mainly @linrock 's machines that are offline. Perhaps someone can contact him on Discord to see what happened?
I am realizing that the double upgrade trick still has the issue that if a worker was offline during the upgrade 241->242 then it can now again not upgrade without...
> sounds a little bit hacky but maybe we can try renaming the lockfile to something else? Great idea! It seems to me this might actually work!
> I am realizing that the double upgrade trick still has the issue that if a worker was offline during the upgrade 241->242 then it can now again not upgrade...
I am sorry for not reacting. I am at a place where I can't seriously think. I will try to reply later tonight.
I think it is indeed best to bail out when the standard algorithm fails, and to use an approximation. The approximation `LLR_normalized_alt()` in the source code would be good.
Your runs do not show worker errors though. I am wondering where the ``` Exception AssertionError at games.py:819 Exception ValueError at games.py:900 ``` come from that the OP is reporting.
This code is the culprit it seems https://github.com/official-stockfish/fishtest/blob/13297926d25a497d4ceefcc9c32bbdfa823e2971/worker/games.py#L895-L905
Yes. But 2. does not show in your runs. However it is reported in the event log. It think sometimes the parsing of the "Score" lines fails. It would be...