UPDATED May 15, 2022
Following extensive technical evaluation of the competitive and training environments during the first three days of the 2022 Mid-Season Invitational, we discovered following Day 3 that there was a discrepancy in the latency being reported in game logs for all matches versus what was being experienced in the Busan venue. To address this issue, we are making a configuration change starting with today’s competition in order to reduce the overall latency to the intended level.
Additionally, because this issue specifically affected teams playing from Busan, the matches involving RNG were held with an unintended disparity in latency between competing teams. As a result, we have determined it is in the best interest of competitive integrity to replay all three of these matches. These replayed games will be finished with all other Group Stage competition by Sunday, May 15; the replacement games will be held beginning tomorrow. We apologize to our players, teams, and fans, and we will continue to work with players to monitor the playing environments in our commitment to maintain the tournament’s competitive integrity.
–Alex Francois, Global Head of Competitive Operations, Riot Games
What has Riot done to date to investigate this latency issue?
On May 11, Riot was made aware of concerns from pro players questioning the latency levels of the competition following the initial games. We heard player feedback and began closely examining the data from our reporting tools. We also enlisted the help of senior Riot engineers from the League Development team who helped deliver Chronobreak and netcode improvements to conduct extensive testing. We looked at server and client logs in an attempt to find any anomalous events or metrics. We also pulled together a structured Q&A sheet and worked with the teams in an attempt to get more information that could help us narrow down where to look for the issue.
On May 12, we set up a test environment where we simulated the network environment from the event. In addition to looking at the networking code, we also started looking at the overall UI event loop (frame rate, game tick code) to see if the latency could possibly be related to something other than the core networking component. This is where we discovered an error in the calculation which was causing a discrepancy between the target latency and the actual latency experienced by players.
Why didn’t Riot catch this sooner? Why wasn’t this tested?
Riot performed extensive testing prior to, and during, the tournament, and our tooling and playtesting showed that there wasn’t a material variance in ping. However, after receiving more detailed player feedback, we designed and implemented custom debugging methods that identified our Latency Tool wasn’t calculating and displaying the correct ping value for players at the venue.
Is the latency reported in game (ctl-F veroverlay) reporting correct numbers?
No, with the latency tool enabled, there is a bug which will result in incorrect reported ping. Players in Busan will have a reported ping that is approximately 13 ms lower than actual ping. So RNG players will see ~37 ms, and Busan players will see ~24 ms, but the actual latency will be equivalent.