Severe Lag after 2000+ players

[Auntie Mangos]

Boyz, you can check MemoryManager patch: http://getmangos.eu/community/showthread.php?p=87171. And test performance of your servers under Windows once again. I know some projects written in C#/.NET, they run only on Windows and they can easily handle 2k+ online

[Kaer]

Hello,

According to our tests on 3.1.3 version of the client about 3 months ago our realm experienced severe (unplayable) latency problems after 2000+ players were online. Other than that, the core is fantastic.

My question is, has the optimization been done to the code since then? Have other servers experienced high latency with 2000+ players online? By the way, our server's hardware is high-end and so is bandwidth.

Cheers

[Clerigon]

our server =

[Kaer]

our server =

Your server lags at 2000+ players? Or doesn't lag?

And which server is it

[Clerigon]

Your server lags at 2000+ players? Or doesn't lag?

And which server is it

the server is spanish

[divisorxx]

You can try newest revs, Ambal make changes that bring more performance to mangos (hopefully he bring more this kind of changes in future). Also you can try patches like mtmaps or visibility refresh times (False.Genesis or SilverIce).

Sorry for my bad english.

[Ambal]

Well, with some custom patches my friends had 4.5k+ ppl online on their Quad Xeon, 16 GB RAM server w/o VMAPS So yeah, we have a progress in performance tuning, just need time to make patches production-ready

[Woweur]

Post your mangosd.conf, and i maybe can help you ( i have 2k players without any lag)

[crashuncle]

Well, with some custom patches my friends had 4.5k+ ppl online on their Quad Xeon, 16 GB RAM server w/o VMAPS So yeah, we have a progress in performance tuning, just need time to make patches production-ready

That is very impressive on such a CPU, but I guess it wasn't very stable yet?

I guess all "testers" connect of course per LAN since anything else wouldn't be allowed *cough*. I personally still use as basic a revision from November08 with Mtmaps and some crashfixes of newer revisions and get with 1000 "testers" around 4-5 days uptime till I mostly restart manually and there is no lag at all. If you don't want lags with the current hardware and Mangos revision then simply set a player limit and keep your database clean. Some members here really demand sometimes "Fix this crash now!" instead of solving it themselves and support the project that way and may even learn something new.

[Ambal]

That is very impressive on such a CPU, but I guess it wasn't very stable yet?

Stability was ~5-6 hours. But it has improved dramatically since we've found the reasons why server crash so often with mtmaps patch... After that 14-16 hours uptime is normal

[xeross155]

Still 14-16 hours uptime is considered weak in normal programming, but I must say, very impressive. I'll contact you whenever I get that many testers.

[voidx]

Ambal your friend use linux or win?

[riddick]

Ambal your friend use linux or win?

I doubt anyone is crazy enough to run 4k users on windows lol.

[davemm]

I doubt anyone is crazy enough to run 4k users on windows lol.

i'd love to see a win machine running anywhere NEAR 4k

[cyrex]

i'd love to see a win machine running anywhere NEAR 4k

My god, that signature is awesome. A Twitter of the changelog. Awesome!

[cryoman]

If you have problems over 2k players, make a max cap of 2k players! The players will fight to enter the realm ... it will be like a VIP server

[cyrex]

Am actually thinking that if you actually got 2k players, then you really dont have any problems. I mean come on, 2000 players in a server is like wayy too many people. If you already got so many people there, then dont worry about lag or spikes. Is like a million dollar man worrying about a dollar bill.

[nanounico]

Hey ambal, help us with the lagg, 2000+ lagg :'(

[NetSky]

i cannot believe you got that bad laggs. With mtmaps and current grid-performance improvements ambal did mangos has the best performance out there in the wide world of wow emus - it is running just like a speedboat *jokingly*

i guess your laggs are mainly caused by a heavy loaded mysqld and bad my.cnf

[kero99]

Stability was ~5-6 hours. But it has improved dramatically since we've found the reasons why server crash so often with mtmaps patch... After that 14-16 hours uptime is normal

Hi Ambal, Can you tell me what is the improvements in mtmaps patch to reduce crash in server please?

We use this patch for rev8582 of kapategit: http://pastebin.ca/1599262

Best Regards ^^

[henhouse]

Good God! Wouldn't that be an internet latency problem? Or are you running it on some dedicated server because that is extreme.

[voidx]

I think using data in item_instance cause mysql lag y big server... because for some reason mangos usin only 1 thread for querys.

[przemratajczak]

yup. I also heard from TC developer that one threaded comunication between DB and core is tendon of Achilles of our project.

[xeross155]

Seperate threads for world and char db seems a good idea.

[voidx]

yes... if your use 1 thread when mysql not return any value server lags....