Desync issues in GFaydark and Felwithe
 

Here goes with the second issue. Once again, I'm pretty newbish at this, so I beg forgiveness in advance. I'm running a server off my own machine, having used the new Windows installer and done a few personal edits (mostly just enabling bots and changing a few respawn times and item stats). I'm running my own login server, but people are also able to connect via the EQemu login. I'm experiencing this issue regardless of which method I myself use to log in. ROF2 client, and I've tried multiple reinstalls from different sources just in case.

Here's what happens: I usually origin back to Kelethin. Originally, for anywhere from one minute to five, nothing in the zone would be responsive. Can't open doors, use merchants, even spells will cast but hang before completing. I can run around and click on things all I like, but until whatever's causing the desync ends, there's nothing I can do client-side to affect it.

Yesterday I had cause to enter Felwithe for the first time, and I noticed the problem was substantially worse. I would desync so bad that I couldn't even zone from Felwithea to Felwitheb, or vice versa. My position was kept (because if I managed to /q out and come back, I'd be where I'd last seen myself, including in the other zone if I was at the zone line), but it seemed like from the moment I logged in it would start desyncing. I was doing the batwing quest and noted how at first it would work fine, then the responses would get slower, and slower, until I was unable to do anything and would have to /q out and wait for the server to boot my character. A few times I had to end EQ via the task manager as not even /q would let me escape.

Sadly, after nearly a full day of trying to figure out what was going wrong, I've been unable to make any difference, and now it seems like Kelethin is getting about as bad as Felwithe. I haven't noticed this problem in other city zones such as Freeport, Qeynos, Neriak, PoKnowledge, etc.

Does this sound like a server issue or a client one? Does it sound like a problem with quad cores? None of the solutions I've tried there have done anything to help it, sadly. I have one friend who has been playing in other zones who hasn't experienced this issue at all, but she hasn't tried to be in the two zones I know are affected. Logging in via my laptop didn't seem to produce this trouble either. I could put up with the original Kelethin annoyance but now both zones are practically unplayable.

Edit: I forgot to add, in case it matters, this machine is running Windows 10, and has an intel processor.

Edit2 to update: I had my friend run around Kelethin while I had been desync'd for quite a while. She had no problems. She could see me, and see me move, but doors wouldn't open on her end either when I clicked on them. She also has a quad-core intel processor and is running Windows 10.

TBH, there's not a lot of confirmed issues with windows 10..so, it will be hard to pinpoint an issue there specifically.


If you can, I would suggest trying another client version - and if possible - trying on an older windows install.

There more possibilities that you can discount, the better the chance of finding a solution.

I'll see about snagging an older client. As for Windows 10, I don't think that's particularly the issue unless something about it would interfere with a client running on the same machine as the server, as my friend is also running Windows 10 and didn't experience any of these issues. Ditto for the RoF2 client; she's also running that one, and no problem.

I know win10 uses the IPv6 '::1' for localhost where older clients use the IPv4 '127.0.0.1'

Not sure if that's the issue..but, you might look for some posts regarding that, if you think packets are not getting through.

I'm afraid I don't really know enough to be able to try anything with the IP6 myself, and I didn't find anything that seemed relevant.

Snagged the Underfoot client, it's having the same issues.

Edit: As another update, I've had on again off again brushes with this issue in the Dreadlands, and Surefall Glade. Occasionally, if the desync resolves itself, I'll get the server message: WARNING: World server connection lost. This only happens if the server and client manage to resync with each other, however. If they don't, it just lets me sit in limbo forever until I /q out.

This is only on your system, running both the game and server?

Correct. If I connect via my laptop (Windows 8, ROF2 client), I haven't run into these issues, though playing on it long term isn't very feasible. Friend with Windows 10, quad-core and the RoF2 client has no issues at all. She was running around killing things and doing quests in Gfay while I was frozen in place for at least thirty minutes without a single slowdown. I'm baffled.

Edit: We can add Erudin Palace to the list of places in which desync seems almost immediate.

I think that you've diagnosed the issue down to playing on the same computer as your server with no zone-specific issues.


If you don't mind, post your 3 config files (eqemu_config.xml, login.ini and eqhost.txt) - just make sure to redact any login information, if you do. (Or, you can pm it to me.)

Sure thing. And thank you muchly for your time with this. :)

EQemu_config.xml

login.ini

And eqhost.ini (note that I've also tried this using login.eqemulator.net, 127.0.0.1, and my public IP, but sadly the issue still pops up).

I generally urge people to use '127.0.0.1' over 'localhost' due to some address resolution issues that can pop-up...

But, since other people are capable of logging in and not experience issues, that's probably not the case here.


Everything looks correct..even where you are logging your client into.

I'll dig around and see if I can find anything on running a client and server on the same computer..of course, this probably won't be eqemulator-specific.

(Maybe someone else has experienced this?)

Went into Howling Stones for the first time on this server and almost immediately got the same desync issue, badly enough that I had to use the task manager to get out of it. In doing so, however, I noticed that three zones (presumably the last three I'd been to), were taking up roughly 30% of the CPU, and a rather fantastic amount of raw memory considering I was the only one on the entire server (my zones are set to linger for thirty minutes or so, so mob corpses don't poof).

Rebooted the server, logged back in, and looked at the task manager again. This is from about twenty seconds having logged into the zone:

http://i.imgur.com/Is20sbN.jpg

My understanding is that EQemu really doesn't need that many resources unless you've got a pretty heavy server load. Is there any reason why it would be hogging this much on a machine that's only a few months old, or am I mistaken? This doesn't really explain why the issues only crop up in certain zones (and instantly at that, with the random exception of Dreadlands), but my CPU load zipped up to 100% several times while I was looking at this. If it's doing that, little wonder my client is getting choked out.

In comparison to the above, standing in Dreadlands for several minutes with a full bot raid up and buffing, the CPU percentage the zone is taking up keeps bouncing between 1-3%. Dreadlands is a zone this happens in, but only infrequently.

Edit: And the desync kicked in just as I was about to zone, so there's another spike up to around 40% CPU load. Again in contrast, killing Gorenaire with a bot raid only kicked it up to 9% at the highest.

Does anyone else who logs into your server use bots in a manner similar to you?

My friend is the only other semi regular player. She's been using a bot group of late (she's still low level), but hasn't had any issues like this with them.

This seems to happen regardless of bots being present too, I'm afraid. When I was testing it last night in Howling Stones, I was logging in and instantly getting desync'd, without even the chance to spawn bots if I wanted to. The same thing happened repeatedly in Felwithe without them as well.

An update on this:

On a whim, I made another account and ran around Felwithe and Gfay. Felwithe has reliably locked up on me within minutes if not immediately. This account, despite being played from the exact same computer, had no issues.

I tried another, older account that was GM flagged (GM not turned on, just the account has high enough status). No problems. Trying to reproduce the effect, I race changed the character to one that can't normally be that class (barbarian/enchanter in this case, versus the one that locks up which is dark elf/ranger). I'm leaving her running in Felwithe for a while, but so far she still hasn't had any issues. At the same time, I logged in my trouble account and had her run around Felwithe. Things desync'd almost immediately. Logged out and back in: same difference. This happened at the exact same time I had the other accounts running around the same zone and interacting with NPCs.

I haven't changed anything with the problem account other than the race change as far as I'm aware. Apart from having played her for a few weeks (during which trouble was noticeable pretty much as soon as I was able to get into Kelethin without being KOS, and from the first time I zoned into Felwithe) I have no idea what's different. I'm going to take a hard look at it to see, but this is somehow even more confusing than before.

Is it possible that the bot data could be affecting the character? I've been raiding with her, so she has something like forty bots. I've discounted bots before since this happens with or without their presence, and because my friend has been able to use them without issue. The only thing I can think of relating to them is if somehow having that amount of bots regardless of being spawned is interacting poorly, which still seems unlikely.

I won't say that bots are definitely out of the question..but, if there were a problem with them, it would most likely crash the server - or at least show up with other players.


I am curious about the #race change, however.

Do your logs show any increase in 'unknown/unhandled opcode' packets with this account compared to accounts that are not experiencing this issue?

This pops up on occasion, but not reliably as to when I'm experience problems:

Just my 2 cents on this from my own experience. I don't run a public server, but my home
server is on a win 7(64 bit) box with 24 GB ram. I was testing out load one day just to see
how much mem/cpu was being used. I started the server with 40 zone launchers and also
logged on 4 different copies of UF client all on the same puter as server. I had no issues
with lag or anything else, but no bots were spawned either.
I have an older computer with 4 GB ram, and that same scenario would be totally hopeless.

So far I've found nothing out of the ordinary with this character or account that I haven't already accounted for, except that, interestingly, the 'hacker' list in the database has flagged it multiple times. Almost every single one of those is in Felwithe, in which it seems to have flagged a number of my disconnects or desyncs as a possible use of macroquest instacamp. It's just a guess, but I assume this is likely the result of my desync-ing and then never re-syncing (and having to either /q out or use the task manager), but it's pretty funny.

Any way you can #race change the affect character back to its original value and test the account?

(Manually changing the database value, in this case, will probably be required.)

I can pretty easily race change her back. Would that mess with her factions?

Also, still tinkering, I copied the character to another account. She appears to have everything but bots. Ran around to every affected area I know about and didn't experience any de-sync. So I flagged the account for GM. No change, things were still good. Then I logged in the affected account, which was camped right next to her. BOTH experienced de-sync after a minute or so. The moment I used task manager to close out the affected account window, the de-sync on the copied character ended.

Is there an easy way to copy over bot data to another character? I'm going to run her around a bit more, but it would be useful to at least thoroughly disprove that the bots are affecting it.

Not real easily...

I made a 'clone' command for 'testing' purposes..but, that only allows the bot owner to clone their own bots.


It could be done with an sql script.

I race changed her (with #permarace, not a DB edit yet) back to the original race (wood elf). Sadly no difference. :(

I get this every time a character is created, but it still creates the character and everything seems to be fine..

versus


This looks like either there's a server query issue - not likely if only one character is affected..

..or, you sent info to the client and it didn't like it and is sending a 'status' packet back that, coincidentally, changed your bind point(s).


Try deleting that bind point entry and see if it corrects itself. (Maybe delete them all for that character..pretty sure there's a default that's set if
no info is found.)


EDIT:

Actually, that looks like the data is being parsed out-of-order.

Those big numbers are probably the float entries - x, y, z and heading - converted to integer.

I think out of order. would that be in the source sending the request to write to the database?

Also, in the character_bind table - any idea what the instance_id is for?

Today I moved over the bots to the copied but seemingly unaffected version of my character (it's on a separate account, this appears to be an account issue since another character on that same account also has the same problem).

Spent a number of hours hitting various raid targets and going to affected zones, and used origin (which takes me to Gfay and pretty much guaranteed a desync previously) at least four times. No issues whatsoever. I'm not exactly going to declare this solved just yet, but it's looking that way.

Now if only I had any idea what caused it. Maybe the account was somehow corrupted? The character and bot data seem to be fine (edit: with the exception of bots not zoning and occasionally getting stuck in bot groups, but that seems to be a separate issue and only showed up recently after an update.)

Do you happen to know the acct id of the affected account?


EDIT:

If you want to start a thread on any particular bot issues, please start it here: http://www.eqemulator.org/forums/forumdisplay.php?f=676 :)

76082

Not one that's ever gone through the official login server though.

Edit: what am I saying, yes it has. I need more sleep, clearly.

Hooboy. Here I am again.

The copied character has been working fine since my last post, no issues, no problems...until yesterday when I got a strange de-sync in Dreadlands. I've noticed a few more since then, and now trying Howling Stones again, it's especially frequent (although so far, the server is resyncing after a minute or so).

Still not a clue what's going on here.

I know there was an issue with Howling Stones at one point..I think you need a key or something to go there or you'll have issues...

There's some old posts regarding that zone around - under Howling Stones or its simile..which I can't remember atm...


Veeshan, I think, had a similar problem too.

Oh, I've got the key, but I'll look around. Thanks again for all your help so far with this. :)

I have had these issues on 7-8 different emus at times usually when using merchant or banker sometimes in inventory it sure is a headache especially if you lose a rare quest piece/valuable item etc

So, an update on this weirdness. I'm still having issues with it in various zones, and it continues to evade my very limited ability to figure out. Strangely enough, between my last post and this one I did a completely clean reinstall of the EQemu database, saved nothing and started fresh. I had no issues for a while, but now I've got back into raiding with bots, it's cropped up again.

Here's the actual reason I'm making this post though: I've found a way to 100% reproduce either this issue or one that's, to me, identical, immediately.

I've been having some issues with zone crashes in Fear. Now and then during a fight things will lock up and I'll have to reboot the client. When this happens, everything in Fear repops except for named mobs. After having it happen to me twice today, I got irritable (I was in there solely to make a go at Cazic), toggled on GM, and depopped the zone, then went to pop Cazic manually.

Every single time I spawn Cazic manually, I get an immediate desync (or crash, if that's what's happening). As with the zone issues, I can move around but can't cast spells, issue commands, etc, until my client finally realizes I'm no longer connected and kicks me. I know this happens as soon as I spawn Cazic because I can see my bots vanish (pets remain, bots disappear, it's weird).

No idea if this helps at all, but there you go. Still using RoF2 client.

Edit: Just for the sake of information, I can have two clients open (one isn't flagged GM), have them standing in the same place, spawn Cazic, and both clients will immediately get the desync.

So just a few questions to iron out somethings and some of these questions are coming from left field lol. How many zones max can your server have running at once? What is your cpu running at for speed? Have you tried disabling bots?

Edit: Looking back I saw the syntax error you posted. I know I was having some desync issues with the same problem before i had to manually fix it in my DB. I think what happens is when you enter certain zones its attempting check those values and its causing an error? Also if it cant write the correct values into the table it might cause a "Loop" in the DB instead of erroring out which is what might cause the excessive cpu use. Not entirely sure but it might be something to take the few mins and fix the error in the DB.

Just noticed the time frame between your last post and this one and it seems you've had this issue for almost a year now ? Am I correct ?

I have a windows box here that I use for testing only, but I do run both the client and test server on the same box, and sometimes two clients.

I'm using a fresh db and code (grabbed yesterday) with no issues, running win 7 64 bit.

Hah im glad you noticed the time stamp I didn't even see the post was a year old lol

Yes, though not straight. There was a big period of time when I wasn't playing (this is just a server I'm running for myself), after which I did the new server reinstall.

Running Windows 10 64 bit, been the same system throughout.