Prepar3D Forums

Chers amis concepteur de P3D,

Voici quelques temps que nous essayons de vous suivre malgré vos multitudes de versions, mais voilà, depuis les version 3.2, nous avons droit à un festival de dysfonctionnements en multijoueurs.
Nous avons testé avec des versions pour développeur, acheté des versions académiques, mais toujours le même résultat au final 20 minutes après le départ groupé de plus de 3 pilotes, les joueurs de désynchronisent et s'éparpillent.
Nous avons chassé "la sorcière" en essayant de débusquer des personnes se connectant avec des versions pirates,
Nous n'avons utilisé que des avions dites de base mais rien à faire.

Ils nous est impossible d'utiliser votre système multijoueur.

Merci d'agir en urgence.


///=========
Dear designer P3D ,

Here are some time as we try to follow you despite your multitudes of versions , but now, since version 3.2 , we are entitled to a malfunction festival in multiplayer .

We tested with versions for developer , bought academic versions , but still the same result in the first 20 minutes after the mass start of more than 3 pilots, out of sync and the players scatter .

We hunted "the witch " by trying to flush out people connecting with pirated versions ,
We used that planes called basic but nothing to do.


It is impossible to use your multiplayer system.

Thank you to act urgently .

Hello there,

Following this news in link above, this issue should be solved
http://www.prepar3d.com/news/2016/06/113614/

But it's not the case , we tried to fly with 4 players under P3D v3.3.5, and the same behavior occurs again, some players become a ghost and keep their last flying datas (heading, speed, altitude) that is make them exit the formation

Is there something we miss ?

Kinds regard

We do confirm same issue as above, 3.3.5 did not corrct the problem,
connecting more than 3 players makes the multiplayer session a nightmare, most of players show to be out of sync, wrong positions showed, ghosts, etc... a nightmare.
Please LM, urgently fix this issue.
cheers
SIM SKUNK WORKS

Hello,

Which type of multiplayer set up are you using (direct IP connect or lobby based discovery)?

Have you tried adjusting the Send Update Intervals section in Additional Options on the host machine? Lower update interval values will give more accurate positioning.

I have uploaded to version 3.4, but the multiplayer function still doesn't work.
After the instructions about the settings of the 6112/6122 , I could make a MP session, but nobody can make the connection to me by direct to my IP adress.
it looks the same problem as in this item,
Joop
PH-MAK

Tested version 3.4.9.18400 in the hope the multiplayer problem was solved even if the changelog did not mention it.
No luck, the problem is still there from version 3.3.x trough 3.4.9, making a session of more than 5 player gets the server crazy and out of sync.
Most of player are barely invisible, other stuck on the air, in few words: UNUSABLE IN MULTIPLAYER.
Reinstalling the 3.2.3 version it returns to work nicely as should be.
We are really disappointed with LM support, at least users deserve a mention and an answer on this problem, instead since September no word about, aside a single post that does not resolved anything.
SSW

SSW,

Cliff received no follow up on his question, we have seen no other reports of the issue, and we have been unable to recreate your issue in our testing environment. Could you provide more detailed reproduction steps?

For example, how many users, what location, what vehicles, how long does it take the issue to appear, etc.? Step by step intructions on what you are doing will give us the best chance to recreate your environment and discover any issues.

Additionally, do any of the users have any add-ons installed? What are they?

There have been multiple reports of some popular add-ons causing major timing issues across multiplayer.

Regards and let us know,
Adam

Thanks for reply Adam,
we can make the best, give you access to our P3D server to use it for tests with all of us,
if you agree we can arrange privately at: info@simskunkworks.com
Meantime i can show you our latest test yesterday night
Number of players: 15
Locations: various from England, Italy, Spain, North Africa
Vehicles: various, most of them SSW airplanes but also stock airplanes
Addons: various, most Orbix and weather engines, also CCP (Carrier and Convoy Planner) creating various airplanes,
CCP works flawless on 3.2.3 while doesn't on 3.4.9
The issue appears after the first connections, first 3-4players have no problems during all sessions, others simply do not see nobody, traffic info gauges shows no players, sometime players see others stuck on air (no position update).
BTW:
i'm guessing why not detach the server from the client ?
I know they run in different processes, is not unusual that even if the client crashes, the server continue to run without flaws until you shut down the crashed client.
I guess the server does not need a graphical interface (aka use of vga) to work, it should be a great advantage, since also virtual servers can be used to run it.
Actually we (VAAFSE www.vaafse.org) are using a dedicated powerfull server (I7 4770 - 32 GB - NVIDIA 4GB - 100 Mb symmetric bandwidth) to host huge session up to 60 players, having a console p3d server we would use a cheaper virtual server.
tia
/Mario

Mario,

To isolate the issue, I'd recommend encouraging your participants to disable all add-ons and use a default aircraft for testing purposes. That will help target where the issue is being triggered. Hopefully this can be done with a small group and be possible for us to recreate on our in house testing network.

Do you have a reproducible test case? For instance, you say everything is fine with 4 players. Does that mean once the 5th player joins, they see no other players? Does that always happen on your setup? What are the vehicles of each player, what are the multiplayer settings, in what order do players join, etc.?

We run extremely large multiplayer instances in house and have yet to see a similar issue, so any insight on how to reproduce it independently will be very helpful in discovering a solution.

Regards,
Adam

Adam Breed wrote:Mario,

To isolate the issue, I'd recommend encouraging your participants to disable all add-ons and use a default aircraft for testing purposes. That will help target where the issue is being triggered. Hopefully this can be done with a small group and be possible for us to recreate on our in house testing network.

Do you have a reproducible test case? For instance, you say everything is fine with 4 players. Does that mean once the 5th player joins, they see no other players? Does that always happen on your setup? What are the vehicles of each player, what are the multiplayer settings, in what order do players join, etc.?

We run extremely large multiplayer instances in house and have yet to see a similar issue, so any insight on how to reproduce it independently will be very helpful in discovering a solution.

Regards,
Adam

Hi Adam,
i can answer to some question here:
- yes we can reproduce steadily the issue, is enough to connect one at time, first 3-4 does not have problems, the following begins to see no players or stuck players, ie: i see a mate when connect but it remains stuck on the position even if is moving .
- our server multiplayer settings are:
* port ranges as usual 6112 trough 6122
* connection timeout 90
* keep alive 15
options on hosting client:
* all checked but "players can use slew mode", "players can pause", "Host can pause"
* voice not enabled, we use teamspeak
* users cannot change realism settings
* all realism setting to max
- We had in the past problems while using narrow band server and huge sessions, now with a wide band (100 Mb symmetric) we did not had problem at all, even if with tenth of players bandwidth occupation never goes over 15%
We will make deep tests in the next days and will report results.

I'm a C/C++ programmer that uses every day the P3D SDK and MSVC 2013, if I may make a suggestion I would say that you could go back from 3.2.3 server that works nicely and see what has changed since then, in my opinion the problem is there, furthermore working to detach server from client could help as well.
tia
/Mario

TEST CASES:
----------------------------
P3D3 version 3.4.9.18400
We did not tested the latest 3.4.14.18870 since not all of us have upgraded and because is unusable until the issue will be fixed.
AIRCRAFT USED: default EXTRA 300, BARON
ADD-ON: almost all disabled
SERVER: a domestic server with wide bandwidth (> 10 Mb)
PLAYERS INVOLVED: 10

INITIAL TEST
---------------------------
We initially attempted a mass connection without a predetermined order.
Server showed to be out of sync and to do not update positions, several players did not see each other, other players were seen in a fixed position while they are moving, others seems to move into a fixed and unwanted direction, others with weird position, nose up or down, etc.

SECOND TEST
------------------------------
A second test was performed connecting one player at time and carefully checking players positions, movement update, control surface movements, etc. before let the subsequent to connect.
No error or out of sync was noticed until the 6th player connected.
As the 6th player connected problems began:
- the last connected player was not seen by all others
- if seen by someone it it was in a fixed position while it was moving
- last connected player was seen assuming weird position
Problems have become more and more serious way with the increase of connected players, basically the situation got back to being as with the mass connection.

OTHER CONSIDERATIONS
----------------------------------------
- it seems that the player who has suffered minor problems was the one who acted as host
- Some player reported that disconnecting after having verified above problems suffered also of a huge frame drop (down to 10 instead our standard of 30)
- We do confirm that version 3.2.3 does not suffer of such problems, they appeared on version 3.3.2 and despite the changelog reported that the problem had been solved this has continued to be in all subsequent versions up to 3.4.9.
We have not tested the 3.4.14 for the reasons mentioned above and also because nothing of what is reported in the changelog, we assumed that the problem was not even addressed.

Hope this helps
Mario
at VAAFSE
http://www.vaafse.org

Mario,

Thanks for investigating further.

Just to confirm - all users were flying the EXTRA 300? Could we get a list of all 6 user aircraft to replicate your setup exactly? Did any users have any add-ons installed? You said they were disabled, but they will need to be removed to rule them out as the cause.

Are you all joining in the lobby or are you having the host start and then having people join after the host has started and is in game? Exact descriptions will ensure we are testing in the same way.

simskunkworks wrote:they appeared on version 3.3.2 and despite the changelog reported that the problem had been solved this has continued to be in all subsequent versions up to 3.4.9.

What changelog report are you referring to?

Regards,
Adam

Adam Breed wrote:Mario,

Thanks for investigating further.

Just to confirm - all users were flying the EXTRA 300? Could we get a list of all 6 user aircraft to replicate your setup exactly? Did any users have any add-ons installed? You said they were disabled, but they will need to be removed to rule them out as the cause.

Are you all joining in the lobby or are you having the host start and then having people join after the host has started and is in game? Exact descriptions will ensure we are testing in the same way.

simskunkworks wrote:they appeared on version 3.3.2 and despite the changelog reported that the problem had been solved this has continued to be in all subsequent versions up to 3.4.9.

What changelog report are you referring to?

Regards,
Adam

- i do confirm 9 extra and 1 baron, first 6 were extra
- addons installed were mostly Orbix and wheather engines (deactivated)
- photoreal (plain bgl's, not deactivated)
really necessary to remove them ? Once disabled and none dll module loaded in memory they are ineffective.
We all are used to have to do with dll.xml and exe.xml , so we can manage to make addon's ineffective.
More over it seems not add-on related since we have different machine setup, someone use REX someone no, most have Orbix, i don't, etc.
If it was caused by a specific addon the issue should not be so generalized, and most important same machine setup works flawlessy on 3.2.3, it seems a clear sign of where the problem lies, just my 2 cents since i don't know how your code has been constructed.
- we had the host start and remote clients connected in sequence as described
- the 3.3.5 changelog:

"Fixed issue that could cause remote multiplayer clients to display incorrectly on slave machines"

- the news:
http://www.prepar3d.com/news/2016/06/113614/
seems to report also something related:

This hotfix corrects two major issues identified quickly after release:
Third party scenery visual anomalies. Various visual artifacts, increased memory usage, and missing features.
Multiplayer sessions only showing other players if they were the last to join.

tia
/Mario