Server failure January 5, 2011

Technical support for DSJ3.
Post Reply
EyTschej
Posts: 279
Joined: 23 Dec 2007, 12:23
Location: Germany
Contact:

Server failure January 5, 2011

Post by EyTschej »

Hey Jussi ... just wanted to tell you that the Mediamond server is like ... totally f-ed today. Major lags and numerous disconnects during the regular Wednesday World Cup competition. I gave up now. Lost quite a lot of points due to disconnects, which is still kinda funny, but you should really investigate this matter. The serverload shouldn't really be that high for DSJ online, so I don't think that about 60 players were too much for the server to handle. This issue needs to be fixed, especially for the upcoming Tour de DSJ, when more than 100 players can be expected to participate.
skill since 1337 ;)
Mein Sport-Blog
Jussi Koskela
Site Admin
Site Admin
Posts: 3637
Joined: 22 Jan 2007, 14:42
Location: Finland
Contact:

Post by Jussi Koskela »

Was it just your connection or other players' too? The server should easily handle 500+ players.
Vid
Posts: 116
Joined: 23 Jan 2007, 03:14
Location: Slovenia
Contact:

Post by Vid »

Many players had lag problems today and got disconnected during the game. Me included, and I have a non-problem running computer.

So it was definitely something wrong with the server.
Who needs drugs when you've got DSJ??
Jussi Koskela
Site Admin
Site Admin
Posts: 3637
Joined: 22 Jan 2007, 14:42
Location: Finland
Contact:

Post by Jussi Koskela »

I can see in the logs that couple of connections timed out during the tournament. As the burden for the server was pretty low I would say that the problem was not the server itself but some international internet connections. Maybe some central european carrier.

In case of server problems all the players should experience equal lagging and disconnecting.
EyTschej
Posts: 279
Joined: 23 Dec 2007, 12:23
Location: Germany
Contact:

Post by EyTschej »

I've never had any connection problems before. Sometimes people talked about "lag" and I didn't believe them, but today I experienced it myself and there was really a high number of players with the same issue. I've tried connecting after the tournament was over in order to see if this had anything to do with the players already connected, but I still couldn't even log on. It simply said "connection to master server timed out" or something.

I've tried tracert www.mediamond.fi just now. I'll leave out the information of the IPs that didn't have any bad results.

1 xxx
2 15 ms 15 ms 15 ms xxx
3 19 ms 16 ms 18 ms xxx
4 21 ms 20 ms 21 ms xxx
5 144 ms 101 ms 20 ms 80.156.160.46
6 68 ms 66 ms 65 ms 67.17.199.254
7 * * * Zeitüberschreitung der Anforderung.
8 * * * Zeitüberschreitung der Anforderung.
9 70 ms 72 ms 67 ms ge-gw1.esp.mediateam.fi [81.17.202.2]
10 68 ms 68 ms 64 ms www.mediamond.fi [81.17.202.111]

As you can see, there is a first delay at #5, but #6 seems fine again. #5 shouldn't cause the trouble though. My bet is definitely on #7 and #8 (timeouts - but no ip/address given).
skill since 1337 ;)
Mein Sport-Blog
Jussi Koskela
Site Admin
Site Admin
Posts: 3637
Joined: 22 Jan 2007, 14:42
Location: Finland
Contact:

Post by Jussi Koskela »

Step #5 (Deutsche Telekom) looks like the culprit to me. There's lots of deviation in those three response times. #7 and #8 might be configured not to respond to tracert at all.
Kajek
Moderator
Posts: 529
Joined: 19 Mar 2008, 22:31
Location: Poland
Contact:

Post by Kajek »

All starting Poles today also had a problem.
http://epgaming.net/ - Jedyny aktywny turniej DSJ3 na świecie!
http://epgaming.net/ - Only active DSJ3 tournament in the world!
Skyflyer
Posts: 64
Joined: 24 Jan 2007, 19:06
Location: Poland, Warsaw
Contact:

Post by Skyflyer »

Yes, most of players had today extremly huge lags. For sure all people from Poland, Germany, Croatia and many many more. In some cases after Qualifikaction jump there were 15 seconds of break with "dead chat and results table" and then 1st round jump, without seeing qualification round results.

The standard duration time of one online World Cup is about 1h15. Today it lasted 1h30m, that means, that 15 minutes has been taken by server lags.

It was really hard to concentrate today - you never know when you will be disconnected. The conditions were awfull today... it can't happen any more in the future! Please Jussi, make something with it. Rivaly in such conditions doesn't make any sense.
JOIN THE BEST DSJ3 TOURNAMENT!
http://www.dsj3.net/index.php?lang=en
Image
Jussi Koskela
Site Admin
Site Admin
Posts: 3637
Joined: 22 Jan 2007, 14:42
Location: Finland
Contact:

Post by Jussi Koskela »

Skyflyer wrote:Yes, most of players had today extremly huge lags. For sure all people from Poland, Germany, Croatia and many many more.
All these could be routed through the same problematic point. If the problem was with a big internet carrier, there's not much I can do.
Jussi Koskela
Site Admin
Site Admin
Posts: 3637
Joined: 22 Jan 2007, 14:42
Location: Finland
Contact:

Post by Jussi Koskela »

Skyflyer wrote:The standard duration time of one online World Cup is about 1h15. Today it lasted 1h30m, that means, that 15 minutes has been taken by server lags.
The duration of the tournament was longer because the server had to wait a bit longer for the results from the players (which it finally then never received).
EyTschej
Posts: 279
Joined: 23 Dec 2007, 12:23
Location: Germany
Contact:

Post by EyTschej »

Deutsche Telekom is a German provider only and shouldn't have to do anything with Poland. In another tracert check, that IP address gave a totally normal response. Still, even with 140ms, there shouldn't be a total disconnect from the server. In times before DSL, pings of 140ms were considered "low" = fast ;)
skill since 1337 ;)
Mein Sport-Blog
Jussi Koskela
Site Admin
Site Admin
Posts: 3637
Joined: 22 Jan 2007, 14:42
Location: Finland
Contact:

Post by Jussi Koskela »

EyTschej wrote:Still, even with 140ms, there shouldn't be a total disconnect from the server. In times before DSL, pings of 140ms were considered "low" = fast ;)
Ping time of 140ms is not a problem, but if three samples give response times that different (144 ms, 101 ms, 20 ms) it looks like that there was some congestion problem which could lead to packet loss. Anyway, at this point it's only speculation where the real problem was and it's probably fixed already. :)
Post Reply