Warfare BE: ACE Edition

[joehunk 10]

Updates!

Version 5b is out. Everyone should update their servers to version 5b as soon as possible. See OP for a detailed change log, but the most important change is that I increased the JIP timeout. It was a 60 seconds, which is about what ACE is supposed to take according to Sickboy. It was timing out and kicking people off even though everything was working properly.

To make sure you don't see the JIP bug, you should run:

- ACE 1.7 pre RC (or later)

- Version 5b of this mission

Anything else runs the risk of seeing the JIP bug. You can get ACE 1.7 pre RC in one of many places including:

- My Addon Sync server (discussed in the original post of this thread)

- Six Updater Suite

- The SR5 Tactical Addon Sync server, auto config URL: http://www.sr5tactical.net/updater/sr5.7z

[Freakboy 10]

I was waiting very long for version 2.066.

Finally we get ki-units spawning in flight.

And the best is KI Teamleaders keep on running!

I tried to add everything you might need plus a little more.

Benny has updated his missions[066] in the evening - So I should too.

[KI Teamleaders now know where to respawn]

Get new version here: http://www.file-upload.net/download-3056374/WarfareV2_066ACE_CO.Takistan.pbo.html

Have fun!

Edited December 17, 2010 by Freakboy
The fundament for this mission has been updated.

[sickboy 13]

I would recommend trying the new OA beta, to see if it helps with the initialization problems etc.

[joehunk 10]

Sickboy, I have not seen any init problems with:

- 5b of my mission, which extends the JIP timeout well beyond the 60 sec you said ACE should legitimately take

- The 1.7 preRC of ACE

So I think it's fixed with this combination, unless I hear differently. JIP still takes a long time, but that's just ACE. The important thing is it works.

This weekend I am going to try my hand at updating my ACE port to Warfare BE 2.066. We'll see how that goes!

[djczing 10]

I still have the same issue with 5b and latest ACE/CBA.

Looking forward to your 2.066 port JoeHunk.

Ran some BE Warfare on Lingor recently -- very nice island - you should take a look.

[wickerman 10]

I would recommend trying the new OA beta, to see if it helps with the initialization problems etc.

The way it looks it is a problem CBA (141 and 144)

a normal BE Server will not work either with CBA

cti has the same problem

sorry for my bad english

[djczing 10]

I think its a combination of things - I noticed the issue starting with arma/CO v1.54.

At the same time Benny (if I remember correctly) changed/optimized things in BE around 2.064 (?).

CBA, of course is always changing, and the new Arma/CO seems to have thrown it a curve-ball in the timing of BE (or perhaps any complex mission) init.

Since CO 1.54, the initialization of BE seems to have broken when CBA is involved. The symptoms are the towns init debug message, and inability to JIP.

I did a little experiment last night and put up BE 2.063 (Vanilla, Chernarus) on my server. This morning, after it had been running about 6 hours I JIP'd - and while the init did take a bit, it was successful. I also JIP'd after about an hour of mission time without a problem. 2.063 also has no problem with Sahrani, where any rev. after that wont init Sahrani towns (for me), either with or without mods.

The bottom line is that when you run CBA (and it is needed by most mods), you will see this issue with BE warfare versions 2.064 and above. Its not the fault of CBA, or BE, or Arma -- its a combination of factors that appear to induce a timing issue.

I know Sickboy is interested and working on the CBA side of it -- Maybe Benny has some insight as to the init changes between 2.063 and 2.064 which would contribute to the issue.

...Or maybe not. This is all very good and free content -- and I would understand completely if we just had to work with what we have -- in which case an ACE port of 2.063 might be an idea, the use of the 2.058 ACE version (Benny used to do ACE ports - and stopped as of 2.058 I think) would be another alternative, althogh losing the excellent upgdates (better AI transportation, helicopters, better AI gear and vehicle purchasing, etc) would be a shame.

Playing with no mods at all isnt really an option for me, as I need to use GL4, Zeus, various sound mods and other CBA-dependent stuff to make the game enjoyable/realistic. ACE is optional for me, but CBA is not.

In any case, I do really appreciate the privelage of being able to benefit from the hard work of Benny, Sickboy, and all the developers of Arma content. Their contributions are immeasurable and have provided this armaholic with countless hours of enjoyment. Whatever can or cant be done to remedy this, I will always have high gratitude for this privelage.

---------- Post added at 12:07 PM ---------- Previous post was at 11:50 AM ----------

By the way -- My test of BE 2.063 was done running CO beta 1.57 client/server and latest ACE and CBA.

[Freakboy 10]

Hey people!

2.066 is running so smooth over here! I can't believe!

Normally I had to reduce amount of slots to 10 or even 8 because of KI

stuttering issues. But now I got all 32 KI in there and it runs better then

2.064 with 16 KIs. So maybe things get better now as server loading in

the moment of joining is a factor influencing init.

[perineum 0]

What does the @CBA_OA part of ACE do? Is it to be used if playing with Operation Arrowhead or Combined Operations?

Any versions of Benny ACE 2.066 for Sara? Or I guess when Joehunk gets this updated to 2.066 will there be a sara version?

[sickboy 13]

Anyone reporting issues with CBA, please find the CBA issue tracker and report with details.

http://dev-heaven.net/projects/cca/issues

JIP still takes a long time, but that's just ACE. The important thing is it works.

ACE is long ready with initialization by the time Warfare is. It must be some combination or condition.

For now I am unable to reproduce the problem, which makes solving it a lot harder.

In my test cases i'm not using signatures, I wonder if it has been reproduced on non signature enabled servers, unsure if it can complicate initialization.

Edited December 20, 2010 by Sickboy

[sickboy 13]

Improved (and signed) Extended EventHandlers available, details at; http://dev-heaven.net/issues/16132

[joehunk 10]

Updated to WarfareBE 2.066!

Version 6b now available from the links in the top post. This version incorporates all the new stuff in WarfareBE 2.066. Enjoy!

As far as the JIP issue, I am not sure what to tell you dj. I ran this new version of the mission overnight for about 8 hours and was able to JIP just fine. Conditions:

- Windows dedicated server

- 1.56

- ACE 1.7 RC1

- 6b version of the mission

I'll try Sickboy's new version to see if it makes JIP faster. Sickboy: you say you haven't been able to reproduce the problem, but have you at least seen that JIP does seem to take a lot longer with ACE (possibly just CBA) than it does with the base game? Depending on what makes it take so long, and how long it makes things take given different conditions, that could be all there is to it...the "0 towns" bug is caused by a timeout.

If you want to try reproducing it, conditions that used to work for me (although I have not tried this recently) are:

- ACE 1.6 stable + the first CBA hot fix (the one that lets you join servers at all in 1.56)

- 1.56

- Warfare BE 2.065 base version

- Windows dedicated server

- Let the mission run until several towns have been capped by your side

- Then try to JIP

Others claim to have seen the problem with later versions of ACE and/or later versions of my Warfare port, but I personally have not.

@Perineum:

I'll try to get a Saralite version out soon.

Edited December 20, 2010 by Joehunk

[sickboy 13]

Had to update due to a possible loop condition; http://dev-heaven.net/issues/16132

@Joehunk:

It is correct that at JIP it seems that all CBA+ACE initialization takes about 5-6 times longer than on ordinary join (from start).

It takes up about 60 seconds on JIP, but much faster if you rejoin (without restarting the game).

This initialization is however during the PreInit state, and therefore runs before the mission (scripts) initializes and should have no effect on the actual initialization of the mission / warfare scripts.

After you enter the game the black screen of Warfare should soon disappear, and probably at least much improved with the updated XEH version available in the ticket.

I will try to repro it with the steps given tnx.

[joehunk 10]

^^ This version is now on my Addon Sync server (see OP for details).

[djczing 10]

Finally got some time to find the RPT files for client/server, and loaded vanilla BE 2.065, with /only/ CBA (no other mods).

The first JIP went fine, second one had the init issue and kicked me out, the third rejoin had the towns init debug msg but fixed itself.

Looking at the client RPT, I see something interesting indicating that something is going on with the towns init (array??) where many towns (more as the number of JIPs goes up) show null objects in the client-side initialization. Nulll objects are usually not a good thing:

[167311,6769.59,0,"XEH: PreInit Started. v3.0.5"]

[167311,6769.61,0,"MISSINIT","warfarev2_065liteco","chernarus",true,false,false]

[167311,6770.19,0,"XEH: PreInit Finished"]

<clip>

"[WFBE (INIT)] Init_Camp: Camp 'MogilevkaCamp3' of town 'Mogilevka' initialization - [Done]"

"[WFBE (INIT)] Init_Camp: Camp 'MogilevkaCamp2' of town 'Mogilevka' initialization - [Done]"

"[WFBE (INIT)] Init_Camp: Camp '<NULL-object>' of town '' initialization - [Done]"

"[WFBE (INIT)] Init_Camp: Camp '<NULL-object>' of town '' initialization - [Done]"

"[WFBE (INIT)] Init_Camp: Camp '<NULL-object>' of town '' initialization - [Done]"

"[WFBE (INIT)] Init_Camp: Camp '<NULL-object>' of town '' initialization - [Done]"

<clip>

Face

Glasses

Face

Glasses

Face

Glasses

Face

Glasses

<clip>

Face

Glasses

Client: Object 2:706 (type Type_268) not found.

[167526,6782.22,6588.42,"CBA_VERSIONING",["#CBA_HASH#",["cba"],[[[0,7,2,144],-1]],[[0,0,0],0]]]

[167526,6783.61,6588.42,"WARNING: PostInit did not finish in a timely fashion"]

Client: Object 2:3436 (type Type_89) not found.

[167528,6783.82,6588.62,"XEH: PostInit Finished; [true,true,true,false,false,true,false,true,true,true]"]

[167557,6784.91,6589.9,"CBA_VERSIONING_SERVER",["#CBA_HASH#",["cba"],[[[0,7,2,144],-1]],[[0,0,0],0]]]

Note the warning that post-init didnt happen in a timely fasion.

Im going to post this in the BE Warfare thread to see what Benny thinks of it -- if need be Ill submit a bug report to devheaven so Sickboy can have a look.

[djczing 10]

Heh - well I dont know how I missed your posts earlier Sickboy -- Im going to test on your new CBA ASAP with non-ACE BE 2.066 (and no other mods).

Ill let you know how it goes.

Im currently running CBA 144 -- looks like you have done some good work on things since.

[sickboy 13]

Thanks djczing, please attach the full RPT file at http://dev-heaven.net/issues/16132

Hopefully it includes some more useful pointers.

Did you restart the game or exit the server completely when rejoining?

Re the null-objects; I am getting those too when joining is proper, basically im getting them always at warfare JIP.

Getting kicked should only happen on team swapping; paramKickTeamswappers

Ah and reading in the Warfare BE thread, there were also improvements to warfare initialization in newer warfare version 2.066.

Edited December 21, 2010 by Sickboy

[djczing 10]

Sickboy - Ill upload both the RPTs from my test above, and also try to do a test on your latest CBA and 2.066 (non-ACE).

Joe - Nice work on 2.066. It might be useful if you could turn on script errors, clear your RPT files and run a mission with some JIPs -- it might be useful for comparison -- although other than the towns init null objects, I didnt see anything else that stuck out in the client RPT, and nothing at all wrong in the server RPT - of course the server side only inits once, at the beginning - but on JIP the server RPT doesnt seem to complain at all.

I think the difference in our server power adds to the argument that timing may be an issue, as im running dedicated server on a quad-core Q6600 - that might push me over the limit as far as timing goes, where you have managed to get back on the right side of the timing issue.

If its just my problem, then I dont want to waste Sickboys time with it -- but theres no reason this shouldnt run correctly on a quad-core with 4gb.

On another note, do you notice in your ACE port that you cant buy some things like HuntIR rounds ? I noticed the same thing on another ACE port of BE (2.064). There are a few items on the menu but you cant buy them.

Thanks to both of you guys for your efforts.

[joehunk 10]

DJ, I will look into the HuntIR rounds. Let me know anything you see where buying doesn't work, or if you buy it and it goes in the wrong slot, whatever. I have already fixed lots of issues along those lines.

[sickboy 13]

Generally I'd sooner appoint the problem to client performance / mods, or server-client connection latency/bandwidth than to the CPU / other hardware.

[djczing 10]

Joe: Thanks - Ill let you know if I see anything else.

Sickboy: Yeah ive thought about that, and have tried to reproduce the issue when server FPS is over 20 (by the end of BE it can get down to 11 FPS, with client side at 20 FPS). Server CPU load stays around 30-40%.

Ive tuned the heck out of both machines (server and client) on the network side, the page file size, etc, set server process priority to real-time and client process priority to high (this actually does give me a few FPS more). Network load/latency shouldnt be an issue, as server monitoring from both #monitor, and from platform monitoring tools indicate relatively little traffic that is nowhere near saturating my 1Gbps LAN. The server config file is tuned for 1Gbps. Ive tested linux and windows servers (currently running windows).

Along those same lines, my thought is that its difficult to seperate whether it might be a loading issue, or a mission complexity issue - does it fail worse as you get further into the game due to the increased complexity of the init, or because the server is loaded down more (likely a little of both) ? Its difficult to answer this question, as its difficult (or impossible) to isolate those 2 conditions.

Oh I forgot to answer a question Sickboy -- in my testing yesterday I did not exit CO and restart - I disconnected from the server and immediately reconnected thinking that having some of this stuff cached would make for quicker load times (and it does) -- but it doesnt seem to effect whether I get the JIP/init issue or not. The number of mods does effect how quickly I can reproduce the problem, as with just CBA loaded I was able to JIP a couple times - and with my normal mod set (CBA, ACE, GL4, Zeus, HiFi, Landtex), or even with just CBA and ACE it will fail JIP every time.

Again, on BE 2.063 with ALL my mods I can JIP just fine.

Benny said he worked some of the null object issues in BE for 2.066 - im interested to test on that version and see how different the client RPT looks after JIP. I did run 2.066 when he released it, with my full mod set, and the JIP issue still did exist. Ill also be using your latest CBA sickboy. Maybe that will cure the problem and all this speculation will be moot !

Joe: After today im on vacation until the new year and should have plenty of time to check into this more. Id be interested to see how my client init goes on your server - so if you can let me know a time when you will be up on 2.066, we can test that out and see what my client RPT says.

Thanks again guys.

[djczing 10]

Sickboy: All I can say is wow. I put your new XEH on both client and server, started the game wiht my full mod set and Joes v 5b version of ACE BE (didnt want to change too many conditions at once).

Ive had it running only an hour, but have rejoined successfully 4 times.

The init is real fast now.

Im gonna let the mission run to completion and keep trying to break it -- but at this point I really think you are on the right track with the new XEH.

Ill let you know tomorrow how the long-run tests work out.

Interestingly, on the initial startup, I got the towns init error from BE, but every JIP has been clean and fast.

Also gonna try it by restarting CO and then rejoining to see if I can break it that way.

Looks like you may have nailed it.

Ill let you know tomorrow.

[sickboy 13]

Very glad to hear, so seems certainly a timing issue :-)

I have continued working on improvements and did some more small ones yesterday

http://dev-heaven.net/issues/16132#note-10

Re the towns init error on first join, was it Takistan? seems mission problem.

Thanks for testing and feedback! Looking forward to hearing results after mission has ran for hours.

Edited December 22, 2010 by Sickboy

[djczing 10]

Well I got up this morning and the OA had crashed on the server.

I was able to test JIP far into the mission last night though.

With towns-amount = full, and about half of the towns capped, and about 8 hours mission run-time the JIP was successful. It did take a bit of time (probably about a minute) where I was standing at the top of the map before the 'Loading (This may take a few minutes) screen came up, but after that it successfully initialized my loadout and spawn.

I did run 2.066/Chux 6B, so it probably wont be too useful for comparison (other than the fact that it WORKED).

I need to back up and run 2.065 vanilla with just CBA I think if you still want to compare RPTs Sickboy.

Further, I appreciate your work in stomping this bug out - its been a tough one -- and id be happy to help you test any interim CBA versions you want.

Or, Ill just get out of everybodys hair and stop taking up bandwidth :)

Ill put up the original RPTs (where it failed). If you want to compare that to an RPT from an updated CBA/XEH, then let me know and Ill run a vanilla test again and upload RPTs.

[sickboy 13]

Thank you :) And NP, with the right feedback and testing we can get a long way :)

Also thanks to JoeHunk and others involved with Warfare BE for ACE :)

Tonight ACE 1.7 RC2 will be released with the latest CBA. Hopefully no new critical issues arise and it will soon be renamed to ACE 1.7 Stable and CBA 0.7.2.

Continued testing and feedback is always useful and welcome :)

Edited December 22, 2010 by Sickboy