Start ADD cowards!
Start ADD cowards!
Almost 90% sure it's happening (on a smaller scale) in NA as well. I was playing with some Texas friends last night (all DFW) and all of us had massive packet loss despite our MS/FPS not changing.
That'd have to be something else entirely - the game servers don't run in the container cluster presently. I didn't see any major network issues with NA last night - possible that y'all are on the same ISP and they were having issues?
You mentioned that a large amount of containers go into making league work, I was wondering just how spread out it actually is? Is a single champ select lobby running on multiple containers at once, each responsible for different things like Champs, runes, Chat, etc? I remember reading about those being highly separate client side, so it would be interesting to know if the server side mirrors the client in that sense.
It's very spread out - the examples you call out in a single champ select are spot on correct. There's a few more but I'm fuzzy on the exact splits so I hope you'll accept a bit of handwave there. In the core game loop we also have other services like the one that manages parties, one for matchmaking, and on the flip side there's separate services for processing mission results and storing your match history, for example. Very much in the microservices style of architecture.
As someone working in the field it is nice to see what happened, what did you do to fix the problem and what is being done to make the chances of it happening again close to zero. I have had my fair share of moments where I've thought the work was done by workers with way too little experience and knowledge to deal with the European servers but I was wrong. Thanks for the article!
We're nearly done moving to a different container scheduler entirely, and that one has much more expressive constraints that will allow us to problematically ensure exclusivity on host across shards with a bit of configuration. Until the edge services can be transitioned over however, we've set up alerting that prompts us to do a manual remap if problematic crowding occurs. That's been working well for the past couple weeks since this incident.
I always like these explanation articles. I may not fully understand everything talked about, but I find it fascinating anyway.
Glad you liked it! If you have any questions, I'm happy to help clarify anything
The power difference isn't that large, but toplane champions are generally useful regardless of the situation. Whereas ADC's are forced to rely upon your team to actually be useful.
Marksmen is the biggest coinflip role & there is nothing to compliment that as not only do you have no agency on the map or hardly any in the lane - you also have a random person in your lane who is 30% of the time autofilled.
'marksmen scale best off gold over gametime' Games don't last long enough for marksmen to get 3+ items
Not to mention the limited amount of marksmen viable right now.
D2 EUW btw
In Elite play (Diamond+) there are currently 9 Marksmen above the 50% winrate line, with all of the rest (except Kai'Sa) being above 48%. I'm not sure if you're conflating viability with playrate? Or maybe just their feel? But if your goal is to win games, you aren't losing them by picking marksmen.
Read moreIgnoring for a second that win rate is well known to be entirely useless as an indicator for power (being about as accurate as a coinflip most of the time), which inexplicably you decided to ignore to make a balance framework that looks at the wrong things, Lee Sin has exceeded 75% pro play presence on several patches, and evaded a nerf. Hell, you have nerfed several significantly weaker jungles, like Shaco, even while Lee Sin dodges the needed nerfs. Lee Sin is dominating in the jungle, and far too powerful.
If all of this is because you're blindly following your balance framework (other than for Akali who qualifies for buffs but isn't getting them), then I think it's time you scrap the current framework and go back to the drawing board. Create a new framework, one that doesn't miss 2/4 of leagues most broken champs of all time, and 3/4 of leagues weakest champs of all time.
Edit: In fact, looking at it, Shaco fails to qualify for any of the frame...
no, winning in league of legends is not a coinflip
Using "in fact" when stating a non-fact detracts severely from your argument and makes people stop bothering with responding to you. especially since in this situation it involves you assuming that i'm lying to you about something you have no way to actually know - so you are already taking your conversation partner in bad faith. hey look, it's shaco crossing the win rate x ban rate threshold
Yeah, but when does that ever come up? I have a million points on anivia and can’t think of a single time where I was like “aw man, if the window for empowered e after being affected by my r was one second longer I would’ve gotten more damage!”
This feels like a placebo that doesn’t address Anivias real problems in that she’s too weak early and never gets to her strong points in most games. Perhaps giving her more roam capabilities or easing her mana issues early so she’s not gated into seraphs+rod of ages?
I don’t know a perfect solution, and maybe this was too much, but I’d like to see some more power given to her so she could have a presence in pro play besides just froggen once a split.
Appreciate the passionate response. I agree Anivia needs some wider work done (maybe QoL, power curve, or power budget shift), but that would require some more due diligence from us that we aren't ready to commit just yet (we swung too far with the CC immunity and were rightly called out for it). We will see what this change does for Anivia (if anything), and reassess after.
How is that pronounced?
Yohn-gi
1 lvl is almost 1k golds so if an Adc is underleveled by 4 lvl + They have bad base stats, can you explain how it’s not as large as we ( all adc players on this planet ) think????
Because the primary output of a marksman is damage (secondary being utility on some marksmen), neither of which are heavily impacted by level scaling, and are instead influenced by gold and base kit utility. Despite being down in levels, marksmen almost always serve their purpose of having the highest damage in the game, which is their most sought after output.
Not without possibility of getting banned
I'm not sure but I know who's the most anti-murican champion: Ziggs
Ziggs and Zilean Jihad squad
Can you elaborate how it is Lee Sin continues to avoid much needed nerfs for months now? The buffs you gave him for worlds (that were completely unneccessary in the first place) pushed him over the edge, and he has been overperforming since, being the best jungler for months now.
Lee Sin sits between 47-50% and ~50% presence in pro play. His play rate is high, but this isn't unusual for a champ that's as popular as he normally, especially not when he is also seeing pro play. Winrate wise, he climbs from weak to totally balanced over elo.
Therefore the only major reason to nerf him would be that we think the game is low on champ diversity since he's too fun - and we don't like punishing champs just because they are fun. You'll notice recently we have been trying to open up the jungle to lots of other popular champs instead (Darius, Garen, Morde, among others).
Is anivia going to get a different buff in 10.7 or is she just going to get the 1 second window that won’t affect anything.
She is getting the 1 second window increase on R for 10.6.
Statistically Amumu's best items are Abyssal Mask, Thornmail, and Liandries, however if you find yourself struggling to stay in range then Rylai's provides a lot of sticking power given that you land your first Q.
As for pathing, Amumu does pretty well when level 3 ganking (especially mid), so I tend to stick to the 3 camp->gank strategy.
Marksmen as a class are less dependent on levels (between stat scaling and skill point scaling) than other classes, and scale best off gold over gametime. So while it may be jarring to see you at a level disadvantage, the actual power difference between you and a toplaner is not as large as you'd think.
Hey! I'd love to give you some pointers sometime. If you're on NA, add me: Riot Ray Yonggi
Yup! Kalista's passive allows her to do a hop in between every auto attack if she issues a move command (right click the ground) once she has started an auto attack.
Yup! Kalista's passive allows her to do a hop in between every auto attack if she issues a move command (right click the ground) once she has started an auto attack.
Just checked. Anivia's Ult is still uncancelable. Trying to get a response from King Cobra to put this to rest.
We've pulled the CC immunity for 10.6. Should be reflected on PBE soon.