Server Issues? Questions? Tips for 4.33.1?

[CobaltUK]

edited

Re my dedicated server query it seems it just does not work well according to Archer so not any solution.

[=VG= SemlerPDX]

Good question - I'll try to contact the original server admin to see why that option is not used. I do know that the Server Pilot has never died, ever - the DPRK never hit Kadena, and if they ever did, they never hit the Server Pilot. So "danger" is not a factor, for sure.

Will post up when I have an answer.

[CobaltUK]

My edit was just too late. Falcon online have found the option just not suitable for a 24/7 server
Host should be in 3D. Host should be first in 3D. Host should be last out of 3D. Host should be last out of the mission UI. 2D server mode is not supported in BMS 4.33.

Also from a Dev ? Scoob_SBM "Host should be in 3D. Host should be first in 3D. Host should be last out of 3D. Host should be last out of the mission UI. 2D server mode is not supported in BMS 4.33. Hopefully that's sufficiently clear and absent of any ambuguity."

As regards the host player does not die ? If so then why does `he` from time to time end up back in 2D ? suggesting that someone/something has crashed into his aircraft parked at kadena

[=VG= SemlerPDX]

Again, good question - but I just have no answer, as I cannot watch it 24/7, no errors are thrown up when it happens, and he is certainly not MIA or KIA when that happens (last time, there was a promotion to Captain window hiding behind the bogus debrief window) so your guess is as good as mine... Also, as we all know, taxiing aircraft have no physical model, AI just drives right through parked planes when it doesn't avoid them (which it usually does)

[Derk]

I saw the VG Viper in the UI on Saturday night around 8 or 9 EST. Thought it was odd, but everything connected and worked for 3 flights. I do continue to have issues with no AWACs comms, even though there were 2 in the local area. I connected at 500 BW, but may try higher as suggested.

As for the VG Viper ending the flight early, I had concern when reading the post about connection to the 4.33 server. Step 4 shows the Veterans Gaming server, highlights in what looks like 'green' to me. Adjacent to that it shows also the 909th wing (Vipers home), and it also looks highlighted green to me (though I know its not). My concern was that someone new to the server (and there are many lately), may accidentally join to that squadron and be messing something up accidentally.

I know it shouldn't do it, but Saturday I actually was able to join in a flight, and was in the same seat as someone else. When I moved to another flight, I could see the other human was there in the seat already. I assume this could happen also if someone accidentally joins the 909th.

Perhaps a strong message about leaving the VG Viper flight alone on the connection information post would be helpful? I'll help keep an eye on the VG host too and let you know if I can come up with anything that could be ending its flight early.

Thanks again for your work keeping things running.

[Brain]

If a flight gets cancelled the client gets the "end flight" window. Considering Viper is online 24/7 an attached timer would automatically cause him to go back to 2D. It's a far stretch but BMS doesn't have a separate EXE for 3D that could CTD like in DCS so it's my only guess.
Why this happens is another question. Usually it's because a flight fails to take off withing a certain time frame, mostly because of a destroyed runway. Unless some troll bombs Kadena every once in a while I have no idea.
Addition: If a player needs to reconnect for some reason, hurries and by accident joins Kadena there is a chance he/she will hijack Viper's plane by accident, as joining a flight through the ATO doesn't change the plane you're sitting in. Since tankers are winchester by default, ending the flight might automatically cause it to be canceled after the player leaves it. We had a similar incident once and it did NOT screw up the original pilot but with BMS you never know.
I was thinking about a separate squadron just for hosting purposes but I think joining a server always defaults to the host's squad anyways.

Lack of AWACS support is not caused by a server bug. Looking at the ATO only one E-3 was actually tasked with AWACS (now called EW/ABCCC, Early warning/Airborne communication command control) ... 2 hours in the future (TO 15:00 server time). The airborne search radars showing on the RWR are E-8 JSTAR planes.

[=VG= SemlerPDX]

The whole problem with the AWACS thing is that it is not dependent of an actual AWACS flight at all! That is a specific option in the config that is NOT set for this very reason - it's a 24/7 Campaign, people fly at all times, and we don't always know that an AWACS flight would be available (or if not, if a new player was playing and didn't know how to frag one) therefore AWACS calls are allowed without needing an AWACS flight.

So this is a rather serious issue. Though it breaks realism to have AWACS always answer as if they were in the air, it's better for the game. So why is it not working? That is the million dollar question.

[Derk]

Thanks Semler. That explains why I see "Chalis 3" on 2d map before and after flight, and can even get a "buddy spike" call from him, but no "Q#" command response. I doubt a million bucks would find the answer, but if I find anything, I'll let you know for free - lol. A to A ops is already dicey without IFF. No AWACs make me shiver and run.

Two flights last night at 1024 BW went fine. Just one odd issue with an AI wingman not following that coincided with a human log off. Possibly his exit burped something as he entered 2d and my wingman spot was all that was open. I have new concerns of these issues. Not sure the best way to hide VG Viper. I've seen others with the default Viper (Joe Pilot) name in the campaign in the past (4.32 ver). Always wondered what that could mess with.

[CobaltUK]

To the best of my recollection the need for a physical AWACS in flight is ON by default. Ive always deselected that since 432. In a big red/blue air war such FO run then destroying the oppositions awacs would be part of the plan but I do not consider it a cheat in the basic campaign where I think they are very safe especially day 2 onwards

[Sinu]

I don't know if anyone else is experiencing this but I'm having a strange issue with weapons employment:

I seem to be unable to actually destroy ground vehicles. I have dropped Rockeyes on trucks, used AGM-65Ds on trucks, hit them with the gun until they exploded and were on fire and when I RTB and go to the briefing screen all I see the disparaging:
"Hit truck with AGM-65 (insert weapon here) - damaged."

I get 0 A-G kills - just damages. Hell, I even used a JSOW on an HART site and got nothing.

In TEs and campaigns I start myself I can slice through the ground units like butter.


@CoblatUK
It is on by default. The first thing I did with Falcon was go into the Config program and uncheck the "AWACS required box" client-side.

[Brain]

I did some testing on AWACS and I found the issue. Offline I created a new campaign and set up different scenarios. The final result is kind of hilarious.

The moment you create your flight and the waypoints show up on the map you are assigned the earliest fragged AWACS plane in the ATO, landing any time after your take-off. It's not his earliest take-off time but literally the order in which the packages were created determines it.
As long as this plane shows up in the squadrons frag order you will have AWACS support, even if the flight will depart 2 hours in the future. It's a small oversight you normally wouldn't notice.
So, as long as the campaign AI keeps fragging at least one of these flights there should be continuous AWACS support ... which leads to the problem: It doesn't!

See...4.33 changed the composition of AWACS packages to include about 6 fighters. Since the majority of F-16s are no longer set to HQ there simply aren't enough planes available to put these packages together!

If I understand Semler correctly he has disabled "AWACS required" and the ultimate question is why it doesn't make a difference. It's a client, not a server setting. AWACS should work 24/7 for everybody who has it unticked in his/her launcher. I obviously had it ticked and spent the better part of the day looking for a bug ... it was kind of fun though :)

@Sinu:
Try recording an ACMI during your attack run. In my case TacViewer showed my weapons only damage some targets but they would blow up just a moment later. Not getting credited for it sure sucks but you're still doing the damage you'd expect and in the end a mission well done is a mission well done!

[Derk]

@CobaltUK- I've unchecked the Config 'AWACS Required' and had success with it for two flights last night. This will be a suitable work around for me, even though I've never experience lost comms to awacs in years flying 4.32 with it checked.

@Semler - everything was great last night except Viper was in 2d (10 est). Upon my return from a flt to Seosan two others had joined. Searching OOB showed 1 at Seosan, 1 at Osan, and 2 at 909th. The others confirmed they were in seats at Osan. When I moved to Osan, the counts changed to 3 at Osan...1 at 909th.

@Sinu- many others and I experience low kill rates in the VG campaign. I assumed it was to allow units to regenerate from damaged state in order to keep up more red forces and lengthen the days of the campaign. If you see "damaged", then you hit it good.

@Brain - Good testing and info. Thanks.

[=VG= SemlerPDX]

Okay - I've double checked the configs again. The server was indeed sitting in the 2D map again, was not KIA or MIA (still 0 deaths in logbook). I've restarted the campaign with a few minor config changes.

I'd advise that all players hoping to help us troubleshoot this and get a good SOP for flying a mission on our server should "uncheck the Config 'AWACS Required' option" in the config.

If it is a full workaround and fixes this issue, we will make it part of the SOP for joining. For now, let's see how this next round of testing goes from this day. Server should at least be stable for 2-3 days, but I am in the middle of a large home construction project and am unavailable to play (my desk is moved out of the way, I can barely play simple games and watch netflix right now)

It's up to you dedicated players to help us out with the feedback to make this server great. Again, thank you all very much for your patience and your input!
!hi

[BigDaddy]

Just wanted to say that I started flying on the VG server a couple of weeks ago and have seen vast improvements on the way that it is set up. It's nice to jump in at any time of the day and catch a flight. It's so much better than flying in single player.
Hats off to all the people that made it possible.

[Sinu]

I got into a circle dogfight with some MiG23's today and I noticed that the AI flight-paths on the server are erratic and seem extremely laggy. I reviewed my ACMI footage from the dogfight and the enemy aircraft appears to teleport around without any ryme or reason. I never bothered to look at other aircraft on external view until this event when I saw both friendly AC and enemy AC seemingly teleporting around or just generally flying against the laws of physics.

I am unsure if this is an indication of a server problem or is just to be expected on any Falcon MP campaign. My missiles were able to destroy the teleporting AI well but going guns on them was nearly impossible.

Here's my ACMI file from the event for those interested with TACVIEW:
https://www.dropbox.com/s/5w9lznkos6r2gev/lag.acmi?dl=0

[Brain]

I think anybody can confirm the AI will do such things. I had F15s chasing a Frogfoot forever at stall speed, using their ABs just to keep airborne and tanks reenacting Tokyo Drift. Interestingly the lead plane (when not engaged) will always behave just fine (AFAIK) but any wingman would be going crazy just to stay in formation.
Shortly after the last restart (Day 1 - about 5:30) I didn't notice any kind of odd behavior at first but soon some MiG-23s (was just observing them, no factor) would start glitching all over the place ....

I dug through a lot of plain text files later, originally looking for a way to change the package composition for AWACS. No luck so far but found something interesting in "atc.ini"! Quote:
This is the amount of time in minutes the player has from the time he enters the sim to get off the ground or get bumped back to the UI (...)
PlayerBumpTime = 50

If this isn't an inactive left over piece of code Viper would be back to UI just 50 minutes after entering 3D. Now that we know 2D hosting isn't supported it might explain why the AI (or BMS in general) will behave so strange after some time:
In 2D you will always see just a flight lead, never any of his wingman no matter how far they stray off...they are counted but don't exist if that makes any sense. When a flight enters a player's bubble the server will still compute the lead plane but struggles with wingman AI as they don't physically exist in 2D. In combat things get worse because the leader isn't simply following his flightplan anymore and combat AI doesn't work properly.

I think we're getting closer to making this server great and stable ;)

[Sinu]

Server unfortunately suffered a blue text "PAUSED" crash today.

[Derk]

I was in on that server crash too around 9:30 PM EST. Did not get the Blue text, but got the windows program has stopped message. Right before this happened I noticed there were two 'Vipers' on the pilot list. I asked in comms who the second Viper was, but no answer. I have concern that two callsigns the same may cause issues with the host. Any thought to giving the VG host a unique callsign.....like "VG Host", instead of using the default 'Viper' name? The VG Viper was in 3D the whole time.

The only other thing I saw at the time is that the server was busy with many clients. I counted at least 8 human pilots on with 6 in flight at once. The server setting of 25,000 BW should have carried all of this with us set at 1024, as long as there is plenty of upload speed on the server side. (about 32 MB/sec upload Speedtest by my calculation)

I did not experience any loss of objects as was reported by two others while in 3D. They said they were set with a BW of 1024, as was my wingman and I. Everything for us worked well up until the crash that most likely knocked everyone offline. Reconnection failed.

Sorry to have to keep reporting issues.
I appreciate your continued effort to provide us a fun/free multiplayer area.

[CobaltUK]

I very much doubt that the multiple use of Viper name is any factor since that's been the case since ever. However ive seen it confuse newcomers to falcon multi play sp why not use `HOST` for better distinction.

I assume the crash logs are examined though probably need some translation by devs

[=VG= SemlerPDX]

edited: The duplicate name possibility of 2 Vipers is not an issue aside from confusing a new player - there is no performance conflict and it's been that way in BMS for years. The callsign has been changed to 'VG SERVER'

I don't use Flash based speed tests for the server, so upload speed tests are expressed in megabits per second (mbps) and are based on the packet size I choose for the test. BMS seems to send 6-9 megabyte (MB) packets if I am interpreting the data correctly (I am NO pro) and the tests are configured for this to give a real world representation of the upload capabilities of the server as they apply to hosting BMS. Results are commonly around 35mbps - if I were to use some generic Flash based speed test, it would give a misleading result of ~65-80mbps.

So many things can interfere with a clients connection to the server, and it's physical location is South Central USA, so international users may experience issues depending on how far they are from there, or the number of hubs they'd need to get there (Aussies, lookin at you - you're the most brave to connect as packets go through Hong Kong before touching West Coast USA)

derk_axelrod LAN_WROTE ...

Sorry to have to keep reporting issues.
I appreciate your continued effort to provide us a fun/free multiplayer area.

No need to apologize - we are so grateful for all of you continuing to fly on the server and reporting here to us!! Even when it's a bit of info that doesn't pan out, or gets ruled out in subsequent discussions, every little bit helps - don't worry about posting up a theory or hypothesis - better for us to put all our minds together on an idea to determine if it's helpful.

The crash last night that was reported saw the server indeed on a crash window, with Win7 asking to terminate the program. So it was 'CTD' instead of 'Crash to Map' and it was restarted to Day 1 with no config changes. I believe it was up for a few days.


I think it might be best to know where everyone is in the world, obviously Cobalt is from the UK.

I'm in West Coast US (so probably have one of the faster connections to the server when I fly).

We should make note of who has dropped textures and how often, as well as how far they are from the server just for the added metric in our testing and understanding of the server's issues.
!hi


EDIT: ***this morning the server was crashed to desktop not 12 hours after the last crash. Restarted to day 1.

[CobaltUK]

Is the server in Kansas ? I do not get a great data flow on the FO Kansas based server.

[=VG= SemlerPDX]

Yes

[Brain]

I'm in GER and never had problems with US based servers (unless some intermediate server fails ofc). Connection is also pretty beefy.

This game is driving me nuts...as soon as you start noticing a pattern and post observations, something happens that doesn't fit in...Broken Mess Simulator 4.33 :D

I noticed there are times when quiet a lot of TEs are going on. Just a few hours ago there were like 5 games with 2+ players each but I didn't encounter any problems. However, it makes me think about "restart BMS to prevent possible memory leaks". What if those TEs don't get "closed" properly and eventually lead to a memory overflow? It's weekend and it's possible there were some TEs going on.
Just some more food for thought.

And just to get it out there: I don't actually know stuff as some might think. I just craft theories based on observations and my basic knowledge about game logic in general. I call it "video game forensics" although I usually prefer really bad games ;)

[Derk]

Brain.... I have to agree that the problems have been random and intermittent.

For me this past weekend, my connections from the north east US have been fine. I have not had issues with missing objects, but I have noticed spinning AI acft and missing landing gear on other human led flights. I did experience a CTD with windows program error message. Two others had it on Sat. night. Reconnecting was fine for all, and the rest of the flt was fine. Was a 4 ship ( east US, Spain, UK, and Denmark). Campaign was early day 1. There were three other TEs running with 6 total users.

At 35 MB/sec UL at the server, the 75% calculator puts you right on for the 25000 BW with the host. How we can run good with all of that running Sat. nigbt, but hear of issues with one user on is beyond my understanding. I'm leaning towards thinking that those with slower or more distant connections are having the issues most. (Australia goes thru Hong Kong? Whoa!). Looking towards that issue, maybe seeing samples of the users Speedtest a Kansas server may help shed some light. Perhaps a separate post just for these specs from users as I've seen on other server forums?

[CobaltUK]

I cant help but think the increased frequency of server crashes is not related to any BW issues and speaking of BW it does seem excessive to multiply the old 432 settings x12 (regardless of any `formula`).

Average number of simultaneous clients is low (very low compared to the FO server) so there is ample bw at any time. Reports of these sort of issues in game when only 1or2 clients, both using 1024 then really adds to the mystery. As does successful no issue flights with 500 bw sharing server with 500 and 1024 players in camp.

Personally at 500 (and probably 300) I have no missing textures other than maybe 1 in 20 flights. What I do have is noticeably frequent A.I wingman spinning on ground or in the air. This can be with other humans on server or none at all. Ive seen spinning A.I on ground whilst humans on ground are positioned correctly and report on radio they see the spinning A.I but that my jet is viewed fine and normal as is theirs to me. Whilst in flight and with a stable A.I wingman ive heard radio chat from players still on ground that they are suffering spinning wingmen.

Ive only seen 1 example of visible lag (jets moving sharply backwards) since 433. Do not know if that is a sign of good data flow to clients for the most part.

For the most part I do not suffer more frequent in flight problems than in 4.32 but there do seem to be more server crashes.