Stablizing Server FPS

[Hi I am John Smith]

Hi all,

Still got no luck on compiling & installing the Kernel.

When booting with the custom Kernel, I am getting the following issue:
-------------------------
insmod: error inserting '/lib/dm-region-hash.ko' : -1 File exists
mount: could not find filesystem '/dev'root' x83
setuproot: moving /dev failed: No such file or directory
setuproot: error mounting /proc: No such file or directory
setuproot: error mounting /sys: No such file or directory
switchroot: mount failed: No such file or directory
Kernel panic - not syncing: Attempted to kill init!
-------------------------

Again, I am using the following:
-------------------------
CentOS v5.4 i386

linux-2.6.31.6.tar.gz (link)
patch-2.6.31.6-rt19.gz (link)
-------------------------

Anyone here have success with them?

It is a bit too much to ask, but if any CentOS guru here can give it a try and report the result? That would be very appreciated.


P.S.
I tried the approach from the official wiki (link) as well, but no luck either...

P.P.S.
As any issues regarding Kernel would end up with a long discussion, can someone start a dedicated thread for discussion on custom Kernel?

People can then report their performance results, and any issues during the compilation and installation.

---

To BehaartesEtwas,

1)
Just install x86_64 on my dedicated box, but I am currently working with my laptop (i386), trying to compile and install the custom Kernel.


2)
Regarding using the idler:
- either on the same core as SRCDS is running on
- or all cores (I am using Q6600)

still getting the same jump as in my above graph.


3)
So for testing, running a single instance of TF2 32 slots, with either no player on the server, or just myself:
- fps_max 0
- Result is ~166

- fps_max 300
- Result is ~200

- fps_max 600
- Result is ~500

yet with any of the above fps_max settings, the actual gameplay performance is laggy. Could it be the network settings on my dedicated box? (I am just using the default)


4)
By the way, what is your suggestion on Hyper-threading? I read some discussion before, that for running SRCDS, it is better to turn it off in BIOS.


Thanks again for your kind replies.

---

(12-07-2009, 10:00 AM)silent Wrote:  The optimization i am using is the one that was located on your wiki page. I downloaded the kernel 2.6.31.6-rt19 and configured it according to the page.

Hi silent, which distro are you using? I am using the same version on CentOS 5.4 i386, but can't even get the custom Kernel to work...

[silent]

(12-07-2009, 10:17 AM)Hi I am John Smith Wrote:  Hi silent, which distro are you using? I am using the same version on CentOS 5.4 i386, but can't even get the custom Kernel to work...

I am using Debian 5.0 x 64.

[BehaartesEtwas]

@silent: what exactly are the symptoms? not pingable after some time of running? does it run well if you use the stock kernel? what version is the stock kernel (give an "uname -a" please)?

@Hi I am John Smith: read the last posts and see what confusion you caused by posting your question into a foreign thread. this does not belong here. and in fact it seems to me not a game server related problem, but a CentOS related problem. I am running that kernel on my gentoo box without problems. I cannot help you installing it on CentOS, I just don't know that linux. I answered question 4) already in that other thread discussing HT... to 3): the only strange is fps_max 0. it should not give anything below some other fps_max value, because it disables all limits. make sure all net settings are correct (see my wiki)

[silent]

The strange things there is no symptoms when i switch to the new kernel it just crash's after a little while of being on. The first time it took a day the second time 2 days before the system became unreachable. When i run the stock kernel the system runs completely fine i was just trying to find a way to stabilize our fps which is has some issues. The stock kernel that i am using is 2.6.26-1-amd64. I can provide any system logs if you want.

[BehaartesEtwas]

(12-07-2009, 08:00 PM)silent Wrote:  The strange things there is no symptoms when i switch to the new kernel it just crash's after a little while of being on.

that is a symptom :-)


and the screen of the server when it crashed, was it blank, or were there the normal stuff written? any reaction to terminal input? (I hope your ISP can check this somehow easily without you paying him much...)

[silent]

(12-08-2009, 06:35 PM)BehaartesEtwas Wrote:  that is a symptom :-)

and the screen of the server when it crashed, was it blank, or were there the normal stuff written? any reaction to terminal input? (I hope your ISP can check this somehow easily without you paying him much...)

They stated that the screen was blank. I don't know if they bothered to try to input anything into the kernel but i am sure they tried.

[Hi I am John Smith]

Hi everyone,

After failing to compile & install a custom kernel on CentOS, I tried doing it with Ubuntu on my box, and everything is fine now.

Now I am going to compile & install a custom kernel for my remote server, and got a quick question:

1)
I am reading this:
http://www.linuxtopia.org/online_books/linux_kernel/kernel_configuration/re147.html

For Q6600, which one should I choose?
- Core 2/newer Xeon
- Generic-x86-64

How about for Q9300, same as Q6600?


Thanks.

P.S.
Regarding CentOS, by default installation, the performance is significantly worse than Debian/Ubuntu. I will come back later to make a post on this.

---

Found this thread:
http://forums.opensuse.org/hardware/64-bit/411439-generic-x86-64-vs-core-2-newer-xeon.html

According to that thread, doesn't matter?

How is your result?

[BehaartesEtwas]

(12-08-2009, 07:57 PM)silent Wrote:  They stated that the screen was blank. I don't know if they bothered to try to input anything into the kernel but i am sure they tried.

hm... maybe you have some console screen saver running. that could prevent a crash message (like a kernel panic) from showing up... sorry, I really don't know how to help you here. those kind of problems have to be debugged systematically. e.g. start with compiling a kernel of the same version of your stock kernel without any patches. then try the version you want to use, still without patches... etc.

@Hi I am John Smith: If you would look at the full name of your CPU, you would know your answer... they are all Core 2 Quad CPUs...

[3choE]

Hi, I am also following this Linux Kernel Optimization guide, and I have question regarding the realtime scheduling part, that whether I should copy and paste the whole script, or there are parts that I need to modify according to my setup?

For example:
1)

Quote:PIDS=`pidof srcds_i686`
for p in $PIDS; do
chrt -f -p 98 $p
done

PIDS=`pidof srcds_i486`
for p in $PIDS; do
chrt -f -p 98 $p
done

PIDS=`pidof srcds_amd`
for p in $PIDS; do
chrt -f -p 98 $p
done

I am using a core 2 quard CPU, so should I only include "pidof srcds_i686", or I need all 3 of those SRCDS entries?

Should I just copy and past that `pidof srcds_...` part, or it should be an unique value for my system? (I am using -pidfile to save the PID of the running SRCDS in file)

Also, I am running 2 instances of SRCDS, do I have to add entry to this script, for each of the 2 instances of SRCDS?

2)

Quote:PIDS=`pidof hlds_i686`
for p in $PIDS; do
chrt -f -p 98 $p
done

PIDS=`pidof hlds_i486`
for p in $PIDS; do
chrt -f -p 98 $p
done

PIDS=`pidof hlds_amd`
for p in $PIDS; do
chrt -f -p 98 $p
done

I am only hosting TF2, so I can take out those 3 pidof hlds ... entries?


Thank you very much.

[BehaartesEtwas]

the script is just fine the way it is. for as many instances as you run. you can take out entries for server types you are not running, but they don't hurt if you keep them.

[3choE]

So basically I can just copy & paste your script, and use it as it is?

Awesome! ( As I am clueless about scripting, so I am happy without messing it up myself )

[BehaartesEtwas]

why don't you just try before asking? in cases where copy&paste doesn't work I have written a warning...