So if I m holding someone at 1 life and not clicking mug . Then someone else will initiate attack against same target and hosp him while he is at 1 life . In normal fight I always see this user is unconsious message which was not followed here.
I am fairly certain that once someone has the leave, mug,etc options that anyone else in attack or joining attack could not get those same options
But with all those clicks happening at the same time, there might just be a moment where someone has finished the attack but not clicked loot yet, and at same instant other players still have the join attack option
my guess is that the scrooge attacks are just different scenario playing out than normal attack scenarios
you have 100s of players (maybe) potentially trying to join attack all at the same time with scrooge
Shows 51 people like the log says and no Hildegarde and the finishing hit. I didn't time out or stale with Scrooge, could you please get us the entire log?
Let me describe what's happened, and how we will handle this. During last year attacking system become faster - because of some optimizations and general servers updates. Also, we now have a lot more active people and also a lot more potential Duke attackers. All this played a bad joke with locking scheme - currently the load is so high, that sometimes it's locking itself into a deadlock. What happens in such case? 2 players landed a final hit at the same time, however because of deadlock they both will be set to the 'somone else finished the fight' status, so there will actually be no winner at all. What we see here is simply a new fight that someone else started after this (because last one already 'finished'), that's why the log with the last winner is so small.
I believe we should track, how it will work now, because changing the locking scheme will require some time and a lot of tests to be done. However, we don't want to close this competition, fortunately we have a way to track, who was the actual winner and credit them an item (this was already done), so all will be fair.
This situation seems to happen only on very high load (we have a way to check, how many times this happened during the last half of the year - and it's zero), in regular group attacks it's almost impossible.
Thanks for your feedback and sorry for inconvenience, such large events are always a challenge, because it's very hard to make a full tests and simulate such activity. But each year we are improving the system, and such events is a great way to find any possible flaws and make it more reliable.