vdbergh
vdbergh
Well it's not a bug AFAICS but expected behavior....
> I can't follow much how all these stuff are related, but I have two concerns: 1- if there can be asynchronous handling to issues that we tackle now with...
To understand this one would need to know how these fleets react to a server restart. Normally a short server restart is transparent to the worker. It should not trigger...
Is there a server log corresponding to this complaint of Sebastronomy. It seems it can have any number of causes. > The fist log is about the issue triggered by...
> > We are routinely taking down some workers fleets, IMO we didn't expect that. > > Because it seemingly doesn't work. This might need a big think for an...
Ok. From the first log. ``` Apr 24 22:26:50 tests.stockfishchess.org pserve[24847]: Request_task: There is already a worker running with name "ChessDBCN-11cores-ccd4f156" which sent an update 70 seconds ago Apr 24...
@ppigazzini Do you see other duplication warnings for this worker "ChessDBCN-11cores-ccd4f156" outside server restarts?
Maybe we should check with @noobpwnftw to make sure he does not have two workers with the name "ChessDBCN-11cores-ccd4f156" (I suspect not since then I think the warning would be...
I suspect that most of these are explainable by people brutally killing the worker, or perhaps by some interaction at server restart. However this I cannot explain: ``` Apr 24...
OMG. It seems `task["last_updated"]` is updated in `tests_view`. That explains it! EDIT: actually not really I think since the change is only done if the field is not present.