Unknown Worlds Forums

Jiriki

Mar 20 2005, 12:15 PM

My server is suffering from high pings. (min. 60-70ms with 16 players and 50ms with 14). Bandiwdth is not a problem, neither is a cpu or mem.

Addons

CODE

Metamod v1.17.4p20 2005/03/05
AMXX 0.20-20050110

Variables (which only? matter)

CODE

"sys_ticrate" is "100.0"
"sv_maxrate" is "50000"
"sv_minrate" is "5000"

Stats:

CODE

CPU In Out Uptime Users FPS Players
90.00 48.81 54.85 5105 1018 19.96 16

Memory

CODE

total used free shared buffers cached
Mem: 504 488 16 0 0 349
-/+ buffers/cache: 138 366
Swap: 0 0 0

Metamod plugins

CODE

description stat pend file vers src load unlod
[ 1] AMX Mod X RUN - amxmodx_mm_i386. v1.00 ini ANY ANY
[ 2] Cheating-Death RUN - cdmod_i686.so v4.21.0 ini Chlvl Chlvl
[ 3] mm_awards RUN - award_mm_i386.so v2.1 ini Chlvl Chlvl

AMXX Plugins (commands.amxx fixed version)

CODE

name version author file status
[ 1] Admin Base (SQL) 1.00 AMXX Dev Team admin_sql.amxx running
[ 2] Admin Commands 1.00 AMXX Dev Team admincmd.amxx running
[ 3] Admin Help 1.00 AMXX Dev Team adminhelp.amxx running
[ 4] Menus Front-End 1.00 AMXX Dev Team menufront.amxx running
[ 5] Commands Menu 1.00 AMXX Dev Team cmdmenu.amxx running
[ 6] Players Menu 1.00 AMXX Dev Team plmenu.amxx running
[ 7] Maps Menu 1.00 AMXX Dev Team mapsmenu.amxx running
[ 8] Admin Chat 1.00 AMXX Dev Team adminchat.amxx running
[ 9] Info. Messages 1.00 AMXX Dev Team imessage.amxx running
[ 10] Admin Votes 1.00 AMXX Dev Team adminvote.amxx running
[ 11] NextMap 0.20 AMXX Dev Team nextmap.amxx running
[ 12] Nextmap Chooser 0.20 AMXX Dev Team mapchooser.amxx running
[ 13] TimeLeft 0.20 AMXX Dev Team timeleft.amxx running
[ 14] Pause Plugins 1.00 AMXX Dev Team pausecfg.amxx running
[ 15] Idle Player Remov 1.00 AMXX Dev Team idlekicker.amxx running
[ 16] NS Commands 1.00 AMXX Dev Team nscommands.amxx running
[ 17] UnStuck 1.00 AMXX Dev Team unstuck.amxx running
[ 18] Simple Slots 1.6.1 DynAstY simpleslots.amx running
18 plugins, 18 running

Because resources aren't depleted, I believe there's some buggy code or so (AMXX's even a nightly build). Another blameful might be SQL, I wonder how optimized it is?

Other than that I'm perfectly satisfied wih my server, so I give a big thank to AMXX/MM devs.

GiGaBiTe

Mar 20 2005, 05:53 PM

90% cpu usage on a 16 man server isnt normal. and i doubt that its a metamod or amxx plugin causing the problem. for this issue, i would look straight at linux. kernels, distros, setup of the linux server.

if you read around, many people have the same problems that you do. i would suggest go bug valve or recompile the kernel or try a different linux distrobution. i prefer fedora because it works stock, but debian, mandrake or suse might work just as well.

cracker_jackmac

Mar 21 2005, 03:30 AM

take a few steps.

Disable Metamod and run a few rounds.
If CPU utilization is still high, then you need a new kernel or hardware.
If not, then disable all the MM plugins and enable one by one.

For best testing, use the same map over and over. because certain maps require more polys than others which play a large effect on CPU (since it is all CPU rendered).

If it turns out that Metamod stuff isn't your problem, there are more than enough threads on linux kernels and distros in these forums to help you out

Depot

Mar 21 2005, 03:37 AM

Definitely ditch the Metamod v1.17.4p20 and replace it with v1.17.4 for a start (it's recommended the "p" version only be used with the patched AMX Mod X v1.01).

I'd ditch the nightly build v0.20 for v1.00 too.

Jiriki

Mar 22 2005, 04:51 PM

QUOTE (Depot @ Mar 21 2005, 06:37 AM)

Definitely ditch the Metamod v1.17.4p20 and replace it with v1.17.4 for a start (it's recommended the "p" version only be used with the patched AMX Mod X v1.01).

I'd ditch the nightly build v0.20 for v1.00 too.

Yeah, I'll try but I hope it's not AMX. I like the nightly-version very much.

I doubt it's a hardware problem, because the server is rented and I got another server with similar configs and some lag too.

Thanks for your answers.

EDIT: Ehm, I 0.2 for v1.00? You mean the nightly build of 1.00?
EDIT2: So I'm installing AMXX 1.01, should I use the patched MM?
EDIT3: MM 1.17.4 + AMX 1.01 gives pings 30-60, cpu is 70-80%. Pingboost is at lvl 3. but fps is at ~50 - weird. Maybe it has something to do with the HZ setting? I have to email admins in order to change that (-> need working value).

Boris

Mar 22 2005, 08:35 PM

My server has no problem with:
1) Metamod v1.17.4p20 2005/03/05
by Will Day
http://www.metamod.org/
Patch: Metamod-P (mm-p) v20
by Jussi Kivilinna
http://koti.mbnet.fi/axh/
compiled: Mar 5 2005, 14:47:15 EET (optimized)

2) AMX Mod X 1.00
Authors: Felix "SniperBeamer" Geyer, David "BAILOPAN" Anderson, Pavol "PM OnoTo" Marko, Jonny "Got His Gun" Bergstrom, and Lukasz "SidLuke" Wlasinski. (http://www.amxmodx.org)
Compiled: Nov 15 2004, 11:17:55
Core mode: JIT

3) updated NS module

I'm running Windows 2003 Server
16 players max (with hidden 17th slot for adminreservation plugin)

I still haven't bothered to try to get AMXX 1.01 to work, last time I did stuff it would just load and then crash without error. I never had the 1.17.4 metamod though...that might help. I have no reason to switch to 1.01 though, I edited all errors I noticed in the admin sma's so everything seems dandy!

Jiriki

Mar 22 2005, 09:16 PM

QUOTE (Boris @ Mar 22 2005, 11:35 PM)

I'm running Windows 2003 Server
16 players max (with hidden 17th slot for adminreservation plugin)

I still haven't bothered to try to get AMXX 1.01 to work, last time I did stuff it would just load and then crash without error. I never had the 1.17.4 metamod though...that might help. I have no reason to switch to 1.01 though, I edited all errors I noticed in the admin sma's so everything seems dandy!

You're running a windows server, that's a different story. Though, I've similar configuration as you (16 slots + 1 hidden). AMXX 0.2-nightly, + MM-p.

Now I have changed to AMXX 1.01 and to MM 1.17.4.

I had a simple problem at start, where it couldn't save log files (because my AMXX was in addons/amxmodx-1.01, not in addons/amxmodx). So I changed directroy paths (the alternative is to edit core.ini). If it doesn't work, you can paste any errors here.

Anyways, to my problem. With only Awards + CD runnig I get only marginally lower pings. Soon I gotta try without MM at all.

I have to test it, but it's our scene's prime server, so you can't play with it all the time.
I have my second server (with similar hardware), but it rarely crowded (because it won't show up with a country flag in the serverlist).

If it goes to my kernel, I'm more than thankful to get any tips. I've heard I should change it to 100, 512 or 1000. Currently it is 1000. I've also heard I should use ck-sources, with no pre-emptible kernel and such. I got such configuration but it's on the second server (which is very hard to get crowded). The other server's kernel won't be changed just like that (because it has to done email).

Note: specs are at top

Jiriki

Apr 22 2005, 09:23 AM

Bump. Anyone? Help really appreciated.

Jiriki

May 20 2005, 06:00 PM

This is strange. I reinstalled NS, changed only a few settings in server.cfg and it still lags. Hlds-tool verified the files are up to date.

Marine_Eater

Jun 1 2005, 11:53 AM

Alreet, would you mind telling us what CPU + MEM you have anyway?

Also, what sys_ticrate are you using, which Linux Distro. + Kernel are you using?

Have you tried Vadakill's kernel tutorial, it DOES work! I went from about 80% usage with sys_ticrate 100 to 18% usage with sys_ticrate 1000 on a ns_siege007 map!

Edit Again:
That's with -pingboost 3 bye the way..

Server Specs:

Intel® Pentium® 4 CPU 2.80GHz
512MB RAM.

Zavaro

Jun 1 2005, 10:53 PM

Never mind.

MarineEater

Jun 2 2005, 08:41 AM

That should have been 'by the way' not 'bye the way'?

Wheres the edit button gone?

Jiriki

Jun 5 2005, 10:06 AM

So AMX, Metamod etc. have little if nothing to do with the lag. Server is rented and the CPU is following.

CODE

processor : 0
vendor_id : AuthenticAMD
cpu family : 6
model : 8
model name : AMD Athlon(TM) XP2400+
stepping : 1
cpu MHz : 2014.851
cache size : 256 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
bogomips : 3973.12

In case you didn't notice, memory-info is here.

CODE

total used free shared buffers cached
Mem: 504 469 34 0 0 326
-/+ buffers/cache: 143 361
Swap: 0 0 0

It means 504 Megs of memory (143M used and 361M free). This was taken when the server had maximum of 16 players.

sys_ticrate is 100. Kernel HZ-value is 1000.
Pingboost has negative effect on pings.
Distro is Gentoo 2003-4.x, not sure which. Where can I check that?

I haven't root access to server so if I have to recompile kernel, I need values which work. Thanks.

MarineEater

Jun 5 2005, 10:25 AM

Do you know if there are any more servers running on that computer?
NS takes up a lot of resources and it doesn't help if there are more servers running on the same computer.

Jiriki

Jun 5 2005, 11:34 AM

Currently ventrilo (basically 0 users) and hltv (0 users too). If you look at the stats, resource usage is pretty low. Hltv is sleeping so it doesnt really matter. There's similar server (same provider) with ventrilo (30-80 users) but it doesn't lag.

CODE

root 1 0.0 0.0 1344 484 ? S 2004 0:40 init [3]
root 2 0.0 0.0 0 0 ? SWN 2004 0:00 [ksoftirqd/0]
root 3 0.0 0.0 0 0 ? SW< 2004 0:00 [events/0]
root 4 0.0 0.0 0 0 ? SW< 2004 0:00 \_ [khelper]
root 5 0.0 0.0 0 0 ? SW< 2004 0:00 \_ [kacpid]
root 6 0.0 0.0 0 0 ? SW< 2004 0:00 \_ [kblockd/0]
root 7 0.0 0.0 0 0 ? SW 2004 0:15 \_ [pdflush]
root 8 0.0 0.0 0 0 ? SW 2004 0:20 \_ [pdflush]
root 10 0.0 0.0 0 0 ? SW< 2004 0:00 \_ [aio/0]
root 9 0.0 0.0 0 0 ? SW 2004 0:17 [kswapd0]
root 11 0.0 0.0 0 0 ? SW 2004 0:00 [kseriod]
root 12 0.0 0.0 0 0 ? SW 2004 12:32 [rpciod]
root 119 0.0 0.5 5664 2928 ? S 2004 0:02 /sbin/devfsd /dev
root 877 0.0 0.1 1664 764 ? S 2004 0:31 /usr/sbin/syslog-ng
bin 987 0.0 0.5 5652 2696 ? S 2004 0:00 /sbin/portmap
root 998 0.0 0.0 0 0 ? SW 2004 0:00 [lockd]
root 1044 0.0 0.1 1608 804 ? S 2004 0:00 /sbin/rpc.statd
ntp 1141 0.0 1.2 6648 6648 ? SL 2004 2:06 /usr/bin/ntpd -p /var/run/ntpd.pid -u ntp:ntp
root 1186 0.0 0.5 5936 2884 ? S 2004 0:19 /usr/sbin/sshd
root 25685 0.0 0.6 8676 3216 ? S 13:07 0:00 \_ sshd: as812 [priv]
as812 25688 0.0 0.6 8956 3380 ? S 13:08 0:00 \_ sshd: as812@pts/3
as812 25689 0.0 0.6 6520 3328 pts/3 S 13:08 0:00 \_ -bash
as812 25921 0.0 0.5 6436 2816 pts/3 R 14:41 0:00 \_ ps auxf
root 1226 0.0 0.5 5640 2720 ? S 2004 0:23 /usr/sbin/cron
as812 5370 0.0 0.1 1968 900 ? S Mar16 0:00 /bin/sh ./run_ventrilo
as812 5371 0.0 0.2 3216 1284 ? S Mar16 0:39 \_ ./ventrilo_srv
as812 25346 0.0 0.1 1968 904 ? S Jun02 0:00 /bin/sh ./run_hltv
as812 25347 8.5 6.6 36216 34264 ? S Jun02 346:36 \_ ./hltv +exec ns/ns-hltv.cfg
as812 25853 0.0 0.1 2232 1024 ? S Jun02 0:00 /bin/sh ./hlds_run -game ns
as812 25630 71.8 15.1 85616 78544 ? S 12:29 90:30 \_ ./hlds_amd -game ns

MarineEater

Jun 19 2005, 10:59 AM

71.8% is a very large CPU usage..