My 1.008 missions redone to stabilize the server

[Sensenmann]

In an effort to help stabilize the server, I have reworked all of my current missions to get them to conform to Lefty's stated requirements per the Out of Memory error.

Plane sets and ground objects have been reduced. Total number of planes in any mission is now at a max of 30, meaning that both recon run missions are limited to 28 players, while the rest are 30 players.

Hopefully this will eliminate any issues with people being dropped mid mission and potentially reduce the lag.

I have updated the version numbers to v10 and already put in the server text (TS info and so forth) so these should be able to be dropped right in to the server without the need for further editing.

Hope it all goes well.

[Sensenmann]

Heh heh...:redface:

Accidentally left my extended mission end timer in place on the Furball mission: instead of the mission ending 30 seconds after completion, it was taking 5 minutes and 30 seconds. So replace the Furball mission in Cooperative2.zip with this one...

[epoch]

Sensenmann

I have no idea if anyone has updated your missions from your first post, let alone your second.

Do you have any idea if your updated missions are running?

If not, I'll address it myself asap.

[Sensenmann]

At this point none of the above files have been downloaded so... nope, still running v9 on all my missions.

Thank you for handling these. :D

[epoch]

Missions added. Server was busy at the time so apologies to those playing at the time.

[Masaq]

Yeah apologies - I'm on nights at the mo, and my connection from my laptop whilst at work is too poor to support access to the server. Thanks for sorting, epoch.

[imp]

it seems like there is still a lot of warping in those missions and eventually everyone gets kicked and the next mission is loaded

[Sensenmann]

Yes, there are a lot of issues with disconnects and lag since the 1.008 patch was released.

I reworked my missions primarily to ensure that no one was getting the "out of memory" error. If it helps with any of the other errors all the better. But the real issue lies with NEOQB optimizing their code for the servers.

From the sounds of it they are working on it. Trying to get the data loads reduced and such. Lets hope it is running better with the release of 1.009.

[imp]

feuerfalke had the theory that it is connected to the AI planes and ground units moving around, do you use them in many missions?

[Sensenmann]

Yes, both Recon Runs have AI flying the recon planes, but there are only two of them. I do have AI manned flak and mg in some of the missions but none of my missions exceed the limits suggested by Lefty at NEOQB.

Those limits are:

Aircraft: Maximum of 6 types and a total of 30

AI Objects: Maximum of 8 types and a total 0f 40

Non-AI Objects: Maximum of 10 types and a total of 30 with Link or 200 without link


Those figure are derived from these statements by Lefty:


1. In a mission there should be not more than 3 different aircraft types for one coalition, i.e. the maximum number of different aircraft types for both sodes - 6.


2. All aircrafts’ skins should be default ones. In this case, application will be using the smallest possible amount of hoster’s computer virtual memory. As an example, every real default skin "eats" about 5Mb of memory, i.e. if a mission has 4 SPADs, each with default skin, it will be equal only to 5Mb of memory, because skins are the same and they are loaded once (in the beginning) and are assigned to all 4 SPADs. Conversely, if these 4 SPADs are ‘dressed’ in different skins, we will need to have more than 20Mb of hoster’s memory to display these four different skins, and the same amount of client’s (player) memory.


3. The maximum total number of aircrafts in a mission at any moment of time shall not exceed 30. This number includes both planes controlled by humans and planes controlled by AI.


4. In addition to aircrafts, there should be not more than 4 different types of ground objects per coalition (cars, trains, antiaircraft guns, cannons, balloons, etc.), i.e. maximum number of different types of ground objects for both sides – 8.


5. The maximum total number of ground objects shall not exceed 40 pieces.


...and further defined by Lefty in response to a few questions I asked him:

1. ...is this restriction placed on AI controlled objects only (vehicles, artillery, etc.)...

Yes only for AI controlled objects.
The others (tents, buildings, etc.):
without Entity there might be up to 200 (not more then 10 different types of them), but with Entity no more then 30 (also 10 different types).


2. ...does the train count as a single object regardless of the carriages added, or is each additional car considered a separate object?...

Yes, as longer train will be as more memory it will use.


Considering there are almost no AI planes flying on the server (Recon Runs the exception) and all other AI is well below these numbers, the problem most likely is not tied to AI.

The #10019 error is known to be tied to the loading times on servers increasing since the 1.008 patch. The problem is the load times exceed the debrief times, so the clients (us) are attempting to load a mission that the server hasn't completed loading.

The #10009 error is an unknown to me. So far it has occurred to me shortly after take off. I am thinking that the server actually crashes at start of the mission and there is a delay in my system recognizing that I am no longer connected to the server, as I am the only craft in the air.

But I have not seen any more of the "out of memory" errors since optimizing all my missions to meet the requirements stated above.

[imp]

Well disregarding what may or may not cause the crahses I can tell that the train raid and the dfw run mission with a relatively full server always have some warping right from the beginning and after 20-30 seconds i get disconnected (server crash?). When that happens all the people on TS also get disconnected.
Earlier today I did actually manage to play the train raid mission but there were only 3 entente and 2 central planes flying yet there also was warping, but less. After a few seconds the warping got less however and eventually stopped. My guess would be that there is either a CPU or bandwitdth usage peak at the beginning of the mission that increases with the number or clients but its merely a guess on my part.

Btw, i did not try to blame you or anything, I just wanted to point out what I had observed. Maybe these two missions could be taken out of the rotation until 1.09 comes out? There are also a few other missions that seem to crash alot, I will post their names when I find out if you like.


edit: "rotary furball dawn v9 wwbrian" also seems to crash

this is not the right thread to post these i suppose, should i make a seperate one?

edit2: maybe there isnt much of a pattern to it since all kind of missions crash but some seem to crash more frequently

[Sensenmann]

No worries, mate, didn't think it was a blame issue, so no offense taken.

And as you stated it does appear to be more of a random thing as the missions seem to crash almost every other one. So a mission will run on one occasion and then could fail the next time it comes around.

I don't know that removing any of the missions is going to solve the issue as it seems to be tied to memory/bandwidth consumption on the server's side. I believe it was Gamecock who suggested on TS tonight that adding a blank interim mission might help purge the server but I m not sure if that is possible.

Hoping 1.009 fixes this issue but knowing NEOQB's two steps forward one step back method of progress, I expect the fixes will come with new problems as always.