Case

If an F-15 armed with slammers gets into a turning fight he deserves to die cause he shouldn't be there in the first place :D

In the 51st we've done things like this but for special events such as GCI weekends or MiG-29A weekends. It does improve teamwork a bit, especially during a GCI weekend when it pays off to be on TeamSpeak. As for normal server operation I'm not sure how useful this will be. If there is a crowd people can start a mission vote to start a new mission, which I think will achieve the same as what you are suggesting.

Just to bump two questions that remain unanswered. master server issues AI behaviour in MP

Will FC3 be released with the fixes to the multiplayer issues people have been reporting in DCS World?

I wish MP would work like that, but it doesn't. As soon as a mission with less F-15's comes up people that can't get a slot leave or start a missionvote for a mission that has more Eagle's. Furthermore, very few people fly for the goals of their side, only for their own goals. You know how people fly, we've discussed this at length in the closing days before FC2 was released. To try and mitigate the unbalance the F-15 brings to FC2 we are running missions with EWR on red and an 80's mission. Both seem more popular compared to the missio. With fewer Eagles.

Remind me to remind you :thumbup:

We certainly will if FC3 is capable of hosting 30+ player multiplayer sessions. I dread to think about adapting my stats software (again!)

Can any beta testers check if this bug still exists? When playing multiplayer in FC2, the AI reacts differently when the human player is the actual server host, or whether he is connected as a client. In the first case the AI reacts like in single player, but in the second case the human player will not get lock warnings and the AI will engage differently (more passive compared to SP). To test this, fly against the AI and compare the difference between joining as server host or as a client.

Here's an update of that plot using over 2 years of data. Bit insane that some missiles were fired almost 30000 times over that period of time. Here is the updated table: +-----------------+------+-------+-------+ | weapon | hits | kills | ratio | +-----------------+------+-------+-------+ | AIM-120B | 2292 | 1387 | 60.5 | | AIM-120C | 8663 | 5321 | 61.4 | | AIM-7M | 1983 | 1451 | 73.2 | | AIM-9M | 1609 | 1072 | 66.6 | | R-27EM (AA-10C) | 1106 | 629 | 56.9 | | R-27ER (AA-10C) | 5113 | 3068 | 60.0 | | R-27ET (AA-10D) | 4473 | 3014 | 67.4 | | R-27R (AA-10A) | 4168 | 2286 | 54.8 | | R-27T (AA-10B) | 2237 | 1360 | 60.8 | | R-60M (AA-8) | 298 | 190 | 63.8 | | R-73 (AA-11) | 4972 | 2975 | 59.8 | | R-77 (AA-12) | 4420 | 2837 | 64.2 | +-----------------+------+-------+-------+ For more interesting statistics; check the 51st stats page

Really? Got a link? :music_whistling:

Yes, as far as I can tell the AI planes stay after kicking the player you think is the hacker.

Do you care to share how you protect your server?

So far the majority of the suspect packets have the same content. I have only seen one other packet that was different. I think though that the problem might be in the header of the packet.

It is not narrow enough yet. PM sent.

It feels like the attacker is least active between 00UTC to 05UTC. You can guess for yourself where on the globe that might place him (after making some assumptions).

Incomplete list of the attacks on the 51st server. Times in UTC+1. 2012-01-27 11:53:55 2012-01-27 11:51:58 2012-01-27 11:50:02 2012-01-27 11:48:05 2012-01-27 11:46:08 2012-01-27 15:52:47 2012-01-28 07:30:45 2012-01-28 07:34:39 2012-01-28 07:36:35 2012-01-28 07:38:32 2012-01-28 07:44:22 2012-01-28 07:46:19 2012-01-28 07:48:16 2012-01-28 07:50:12 2012-01-28 07:52:09 2012-01-28 07:54:06 2012-01-28 14:55:48 2012-01-28 15:39:56 2012-01-28 15:40:38 2012-01-28 22:05:35 2012-01-29 19:09:52 2012-01-29 21:03:20 2012-01-29 22:21:48 2012-01-29 22:43:03 2012-01-30 03:13:53 2012-01-30 09:50:03 2012-01-30 18:45:22 2012-01-30 18:50:54 2012-01-30 20:20:06 2012-01-30 20:20:21 2012-01-30 20:21:07 2012-02-01 12:07:29 2012-02-02 09:08:11 This list is biased as I can not always put the server immediately back online (and get fed up with doing it as well). Most of the attacks which are spaced closely in time originate from the same IP. Some of the IPs of attacks spaced by more than that do show similarities. I can share the IPs on request.

No, passwording the server does not help. The hacker only has to send a specific packet to the right IP and port, and regardless if the server is passworded, it will crash.

Thanks for the information EB1. If this can not be implemented in FC2, would ED be willing to explain to server admins how these attacks bring down a server? This may help us protect our servers.

and only works for HTTP requests, not ordinary TCP and UDP network packets...

I think it is done manually, where he reads the connections from the server list and then attacks through a proxy. I have repeatedly changed port and each time the server was brought down from the proxied IP that IP only had made one connection with the server, i.e. the attack is specific to the present IP and port and he is not attacking all ports.

I can show you a similar screenshot with only 3 servers. Honestly, I can assure you that the problem is real and we are not imagining it.

it would help if ED could inform server admins what to look for. I can identify the packet that crash the server based on their content, but I'd like to know anything that can help us stop these attacks.

PM's between 51st members and Dmut, USSR Rik and GG.

Apparently ED has fixed this for FC3 but will not patch FC2/BS1. Unfortunately they have not provided any information to server admins on how to prevent these crashes... :angry:

That depends how it works. It would be impossible to route all traffic through the master server, which means that clients will only use the master server to find a server, but then all traffic would be direct. Chances are that if that is the case, and this bug is not fixed, FC3 could suffer from this as well.