15 replies to this topic

[Pleased]

Everytime when they are like 50 people or more in Generals/ZeroHour

or when they are 150+ Users logged in the whole system the server is shutting down.

Sometimes people getting knocked out, are having lag/latency problems.

 

Is this caused by the number of users or is this just a coincidence?

 

 

Im not begging for support or whining about the situation but it would be interesting to know why this happens.

 

[The_fury]

I was playing coop earlier. As soon as the mission ended, we both got disconnected and can't login again.

[Graion Dilach]

You're right, the server can't really handle this currently. There are already some hosted projects which went completely broken just due to CnC Online.

Edited by Graion Dilach, 18 August 2014 - 12:20 PM.

[Mathijs]

You're right, the server can't really handle this currently. There are already some hosted projects which went completely broken just due to CnC Online.

What do you mean?

[Banshee]

Several sites in the server, including PPM and CNCNZ are being haunted with error 503s (service unavailable) recently. There is something overloading the server and there is a high probability of being CnC: Online.

[Mathijs]

I really don't think that's possible. The main Revora site/forums, the C&C:O and T3A:O websites don't suffer from this. Pretty sure the multiplayer bit of the server is seperate.

[Banshee]

The error 503 happens with accounts that makes a higher usage of hard disk. The problem has raised with the high activity and constant errors at the C&C: Online, which reduces the availability of the server I/O resources (probably due to services being restarted, logs being written, etc). In PPM's case, the problem is that it handles many sites in a single account, with some popular projects such as Twisted Insurrection and The Dawn of the Tiberium Age. That's why it suffers more error 503s than Revora's main account. I'll tone down this issue in the next days. CNCNZ's case is slightly more complicated, because it is very active and it uses some heavy softwares. But I think that, even if I move some sites out of PPM's main account, we'll still have a lot of error 503s in several hostees if C&C: Online continues to be unstable and requires a lot of restarts.

[ICT]

Yes, all together is no long term solution. But maybe further details on this matter should be discussed somewhere else.

[Banshee]

Of course that it would be better if C&C: Online was in another server, but I think the current server can handle once the C&C: Online software gets more mature, specially if the next step for that to happen gets actually properly tested and transferred to this server.

I know there are better places to discuss these things, but, sometimes, a small public pressure is necessary and it won't harm anyone.

[ICT]

My thought is that certain people could miss the discussion/information down here

[Mathijs]

And that those who do see it, aren't in a position to do much

[Banshee]

They are in position to test the dev server and invite more people to help testing it .

[Mathijs]

Not when the dev server is down and we don't have the keys to restart it...

[Banshee]

Oh, true. In this case, things really get much more complicated.

[xezon]

Can someone profile the server and find where the bottleneck is? Is Server CPU hitting 100%? Or is there a memory leak that screws the RAM usage, and finally fails to allocate memory?

[Banshee]

Error 503 is usually related to hard disk usage, instead of CPU and RAM. However, high hard disk usage can happen due to the usage of swap memory. So, it's possibly related to RAM.