The Diablo 2 Resurrected server status issue was one of the biggest this week. Reports of login issues and loss of character progression have been coming in for days and players have been rightly annoyed by the constant posts on Twitter. Well, there is one Longest statement about server problem.
D2R: How did the disaster happen?
First of all, Diablo Community Manager PezRadar explains that server failures are not a single, recurring problem. There are basically big and small problems, some of which are mutually dependent, and at the moment Blizzard has no choice but to post bug fixes for each issue individually and see if they solve the problem overall. At the same time, you are also working on long-term architectural changes, and here the so-called legacy code causes some difficulties – but more on that later. According to Blizzard, a “small number of players” suffered a loss of character progress – in the future, any loss due to server downtime should be limited to “a few minutes”.
In order to better understand the issues that have severely affected the fun of Diablo 2 Resurrected since Saturday, Blizzard explains the approximate structure of server databases. The center represents the global database that stores all the information and presents the characters. Regional databases (European Union, NA, Asia) act as support, also for low response times, that manage information and profile at the regional level and regularly transfer this data to the global database. Most of the player’s actions are carried out using regional databases, as it is faster and the character is almost unlocked there in order to preserve the integrity of the individual character’s data.
D2R: Drama from Saturday to Tuesday
Below, Blizzard explains the chain of events that led to the usual server problems since Saturday. The starting point was Saturday’s “sudden surge in traffic” that reached a level not even reached at launch. It was made worse by Friday’s update, which was actually meant to improve game build performance. However, both together caused an overload on the global database and the corresponding timeout. Then it was decided to undo the patch to relieve the server. As a result, there was actually higher traffic and too many connection attempts from the game servers. “Hundreds of thousands of games in ten minutes” had overloaded the database again.
On Monday, they wanted to import configuration and code improvements to the global backup database. This led to another outage where the backup database accidentally continued to perform the backup process instead of processing requests from the servers. In the background, Blizzard was working on several improvements and was also sure that they had at least fixed the large number of lost connections to the database permanently. However, on Tuesday there was another large number of simultaneous players, with “a few hundred thousand players in just one region”. This further degrades the performance of the database, which database engineers are investigating the cause of this. Help has also been sought from other Blizzard teams.
“Social media evangelist. Baconaholic. Devoted reader. Twitter scholar. Avid coffee trailblazer.”