starting a GSP (Game Server Provider) within the next 4-6 months.
One big factor I will need to tackle is optimizing the linux kernel on CentOS for game servers.
This entails accelerating the kernel to allow for 1000fps in Counter strike and Counter Strike Source, as well as tailoring the kernel to provide very reliable server side fps.
My question is, what would be some good sources about possibly doing this, also how would I go about rebuilding the kernel, and is it worth learning how to do myself even though my knowledge is fairly minimal in Linux?
Also does anyone know of specifics about what I need to modify in order to do this.
Programs compiled with the Intel C++ compiler run 20-30% (up to 450% in some rare cases) faster than their GCC counterparts.
I have successfully compiled MySQL 5, PHP 5, eAccelerator and nginx with ICC 10.1.015. I was wondering, if anyone has compiled the linux kernel with icc? There are some linux vendors who supply icc compiled kernels, and there are some docs on intels site about compiling the kernel with icc... so it's doable.
But I want to know if anyone has done it on a VPS? What was your experience? Did the system run faster than before? Or did you face some weird situations?
Fedora Core 6 In a first step, the yum repository must be configured. The .repo file is available in Ingo Molnar's project directory [url]. # cd /etc/yum.repos.d # wget http://people.redhat.com/mingo/realtime-preempt/rt.repo Only the first time, yum must be called in installation mode: # yum install kernel-rt Later on, the command # yum update kernel-rt will update the kernel patch, should a more recent version be available. By default, the newly installed kernel is made the default boot kernel. The realtime-enabled kernel is, therefore, immediately active after the system has been restarted. To check if enabled type: # cat /proc/sys/kernel/kernel_preemption 1 Then this: # cat /proc/sys/kernel/preempt_max_latency 39 As an debugging aid, the condition that led to this latency is also available: # cat /proc/latency_trace preemption latency trace v1.1.5 on 2.6.19-1.rt10.0001 -------------------------------------------------------------------- latency: 39 us, #2/2, CPU#0 | (M:rt VP:0, KP:0, SP:1 HP:1 #P:1) ----------------- | task: posix_cpu_timer-3 (uid:0 nice:0 policy:1 rt_prio:99) ----------------- _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | caller / ||||| | / <...>-3 0...1 39us : __schedule (__schedule) As a role of thumb, the maximum (worst-case) latency amounts to approximately 105 / clock frequency on an idle system. For example, a maximum latency of about 100 microseconds can be expected in a 1 GHz-CPU. If the value displayed is off by more than one order of magnitude, something is not working correctly.
Totally lost here! can someone explain this latency figure of 39, how do I know if this is good or bad figure?
iowait creates high performance degradation. There was no changes in server software & configuration. Sometimes server work very slow, count of apache processes is high - there is no any processes and scripts who can use a lot of hdd resources.
Sites located on customer server does not create high load, and not subjected to any attacks.
I received response from DC:
I did some research for this and it seems that other people has experienced similar issues and is reporting that it is a kernel bug have you ran yum update to receive the latest kernel if not you can try and update and see if that may resolve your issue.
CENTOS 5.2 Cpanel / WHM Linux *** 2.6.18-92.1.22.el5PAE #1 SMP Tue Dec 16 12:36:25 EST 2008 i686 i686 i386 GNU/Linux
I try to run "yum update kernel" but it seems installed kernel is up to date.
I am currently researching the options open to me for Virtualisation, the two main ones I have seen are Xen or KVM.
I mainly use CentOS (RHEL), but have read that the version of Xen with it is very old, broken and unstable. KVM isn't included in the kernel that ships with CentOS, as it is too old, apparently it was first featured in Kernel v2.6.20. There isn't likely to be an update till RHEL6, which is due for release first quarter of 2010. I can't wait over a year, so need to find another Distro for use as the Host OS/Hypervisor.
I have built a pretty powerful server, it has an Intel Xeon 3230 which has VT - so I might be better off using KVM over Xen. I am going to collocate this server, so realistically I can make this decision only once - as it would be a PITA to re-install a host Linux distro remotely.
I did a search on distrowatch for distros with the latest version of the kernel, and Slackware came up as being just one minor version behind the most current (v2.6.27.7).
Now this distro is very mature, so should be a fairly safe bet, but it is a 32bit version and can't host 64bit VMs. I have 8GB of ram so want to be able to use it all, and offer the choice of 32/64bit VMs. So that's that out of the window.
I have used Arch Linux on and off for a couple of years as a workstation OS, but because it is so bleeding edge, when pacman updates it can break itself. But I suppose if I just use it as the Host OS, and never let it update/reboot, then it won't break. It should be fairly lightweight and stable, as I will be installing the bare minimum packages. I have a management card, so if the server fails to boot, then I can still remote in to fix it.
If I do want to update the kernel, is it possible to update without rebooting? I think it is somehow... unless I can just reboot during an unused time at 3am or something.
As you can tell I am leaning towards KVM on Arch Linux (x86_64). Is this a good plan?
Is it safe to upgrade to the latest Linux kernel version 2.6.19.2 (released on January 10th, 2007) ? Is there any reported problems or have anyone faced issues after upgrading?
we have one box in hivelocity.net that has been down so many times this month that we were forced to remove links to siteuptime where we were once so proud of having a 99.7% uptime for 3 years in theplanet.
syslog shows that just before crashing, these entries were made:
kernel: kernel BUG at mm/rmap.c:479 kernel: invalid operand:0000 [#1]
dmesg also shows this:
... Brought up 2 CPUs zapping low mappings. checking if image is initramfs... it is Freeing initrd memory: 482k freed NET: Registered protocol family 16 PCI: PCI BIOS revision 2.10 entry at 0xf9f20, last bus=1 PCI: Using configuration type 1 mtrr: v2.0 (20020519) mtrr: your CPUs had inconsistent fixed MTRR settings mtrr: probably your BIOS does not setup all CPUs. mtrr: corrected configuration. ...
i've googled these messages and they point to ram problems.
hivelocity.net claims to have done diagnostics on the box and that there were no problems reported.
they said this is a result of a sys configuration problem made by us.
Last year I ordered a new server with Centos 4.3 and it had the kernel kernel 2.6.9-34.0.2ELsmp installed. It runned fine and I didn't update any packages since then.
Today I started getting a problem where both mysqld and kswapd0 uses very high amounts of CPU, spiking up to 100% and my memory usage is at 99% all the time. The problem seems exactly the same as the one mentioned in this thread.
In that thread the exact same kernel is said to be insecure and to cause this problem. I also came across a centOS bug that reports this problem with high cpu, mem usage and mysql & kswapd0 consuming all resources.
In the linked thread the person solved the problem by upgrading to kernel 2.6.9-42 using rpms but others recommended a newer kernel or a custom compiled kernel for CentOS.
Apparently when they used yum it said 34.0.2 was the latest kernel.
What should I do to upgrade the kernel, which version should i upgrade to, and where do I get it from? I won't be able to compile a custom kernel and I've only installed basic rpm packages before.
I'm running a Xeon Quad Core 2.5GHz with 2GB RAM. Comments on my setup for shared hosting? (will upgrade ram if people believe that will help with increase page loadtimes.)
Warning: mysql_connect() [function.mysql-connect]: Can't create a new thread (errno 11); if you are not out of available memory, you can consult the manual for a possible OS-dependent bug in /home/user/public_html/file.php on line 400 Can't create a new thread (errno 11); if you are not out of available memory, you can consult the manual for a possible OS-dependent bug
currently im on a VPS server and things are fine normally. However whenever there are like 20 users on my site (browser based text game) things get ugly. The sql queries are all needed and don't think it can get much more effiecent.
Anyways my website is very dynamic all information is changing every second.
For example a user when every page they go to or refresh their account is being set to time() to show that they are online at that second. Processor load gets up to like 6 at times and that is when it is virtually impossible to connect to the site.
Is there anythign you recommend? mysql is the top then when i login via ssh and run 'top'
I have a large proxy, are getting ~3k pageviews/day.
I only got 128 MB RAM on my VPS, so no panels works on it (not even kloxo/lxa). I would guess apache wont do it in the long run either.
So I decided to go with nginx. But I feel it isn't quite good optimized, as it sometimes takes like 1 minute to load the page.
Here is my current config:
####################################################################### # # This is the main Nginx configuration file. # # More information about the configuration options is available on # * the English wiki - http://wiki.codemongers.com/Main # * the Russian documentation - http://sysoev.ru/nginx/ # #######################################################################
#---------------------------------------------------------------------- # Main Module - directives that cover basic functionality # # http://wiki.codemongers.com/NginxMainModule # #----------------------------------------------------------------------
I am facing problem regarding my site. I have dedicated server with WHM/cPanel, i have integrated tomcat with apache , the apache processes are too high aroung 1200-1400, So please guide me howto optimize my apache for better performance.
root@trick [/]# ps aux | grep -c httpd 1403 root@trick [/]# ps aux | grep -c mysql 3 root@trick [/l]# free -m total used free shared buffers cached Mem: 4053 3408 645 0 220 1890 -/+ buffers/cache: 1297 2756 Swap: 4094 0 4094
---------------------------------------------
KeepAlive On MaxKeepAliveRequests 100 KeepAliveTimeout 15
I'm looking to get some tips to optimize my Apache webservers. Now i have 2 apache servers with load balancer. If someone can guide me and bring some tips for this purpose is highly appreciated.
Also, i would like to know about disadvantages of enabling KeepAlive. Recently i have enabled it and found my site load much better. But the disadvantage (that i experienced personally) is after stop browsing for a while (about 40-60 minutes) i need to keep triggering the URL of site several times to get enter the website. If not i can't get into it and it likes the site giving no response at all.
right ive got a server thats getting battered at the moment by lots of apache processes, ive moved sql off to a seperate box (with private lan so thats fine) but now i need to optimize/replace apache with something less intensive on resources, only problem is im using plesk on this server so i am not sure if i can change the webserver app?
I am switching from shared hosting to managed VPS. I have minimal knowledge of UNIX/Apache, pretty good with PHP and programming in general.
So, what I have now is my VPS, with 512mb of ram, CentOS, cpanel and my two sites, not live yet, that is, I haven't changed dns records in the registrar.
I want to try to optimize the box or whatever is the correct name for it and understand the whole thing better. Mainly to learn more, I dot' think I'll be pressed for memory or bandwidth for now. Still, at least knowing things I can optimize for speed of space.
Few things that I assume should allow for some simplification and RAM savings(hopefully) is that:
a) Both sites are hobby sites, not very big load. Few thousand visitors a day. Traffic is around 10gb monthly combined.
b) Both sites are mine, nobody else on that box, so I don't have to worry about customers screwing things up and caging things internally.
Based on that, I don't really need email for the second site domain.
When I run top command I see 2 exim processes running. Is that normal, or is it because I have 2 domains registered there? And if yes can one be disabled?
Same applies to pop3-login and imap-login processes, 2 copies of each. Can I have just one of each, i.e. exim, pop3-login and imap-login?
Next, I selected not to run log analyzer cpanellogd for 21 hours, i.e. it'll run for 3 hours a day. I assume that should be sufficient for the site like mine?
I did read sticky about securing the site and quite a bit about Apache/mysql optimizations.
I've got a VPS server with a total process limit of 100 process.
I know this sounds like a newbie question, what numbers would you recommend for the Apache 2 Conf file to get it running as best as possible without it taking the server down?
i am having quite problem with my server i cant get it to optimize it. During peak times the load goes beyond 10 sometimes 20 to 50 sometimes its just under 1. I have my webserver running on P4 Core 2 6300, 1 GB ram, 250 HDD, on CentOS 4.4 (kernel 2.6.9-42.0.10) etc. It has cpanel/whm running Apache 1.3.7, php 4.4, mysql 4.1, mailscanner, eaccelerator, etc i dont know really what other info to show you.
I host about 200 accounts and there are 4 large php forums (not like WHT) running on it. Sometimes pages stop loading especially these forums have to refresh to get page load.
During day there are approx 700-900 connections.
Previously i had phpsuexec and the load never went beyond 4-5, i have removed it to install eaccelerator.
My troubles started about a month ago, on a Windows 2003 VPS (clean, no plesk/cpanel): Every ting worked fine until, for no obvious reason, IIS started to consume too much memory, I tried separating sites (two) in separate application pools, adjusting recycle triggers - nothing worked, sooner or later worker processes would use too much memory and fail, only way to fix the problem was to restart it.
At this point I would like to mention that: - MySQL is optimized for low memory usage (and task manager never said otherwise) - PHP is limited to 20MB RAM, only few modules were enabled - the only other application running is WinDNS, and it doesn't seam RAM hungry
So a few days ago I decided to kill IIS/app server role altogether and install apache, something I should have done long time ago. Everything went smooth, VPS performance dramatically increased, memory usage was negotiable... at first.
As time goes by, apache consumed more memory (httpd.exe comes to about 80MB in 12 hours), it doesn't seam to be dependant on site traffic, and it doesn't drop, server didn't fail once yet and I have apache on scheduled graceful restarts every 24hrs.
I didn't try setting a MaxRequestsPerChild value, I'll keep that as a last option because I really want to get to the bottom of this
So... I suspect it could be PHP, the only thing that's common in both scenarios. I did upgrade to the latest version during Apache installation though.
if using the Internap FCP technology to optimize the routing. I want a feedback on it, since I want to deploy this solution for have a better traffict routing.
Also, is anyone using avaya? i have looked in their website, but I have no information about their routing optimizer. Basically I want to go beyond of normal BGP since I will be deploying VoIP services soon.
just got a new server and got my site up and running on it perfectly but ive noticed its a bit slower than my old server and i think this is down to my incorrctly optimized mysql setup. ive used the my.cnf off my old server with some minor edits but im wondering what people can suggest i change for the following server spec
AMD Athlon 64 X2 4200+ 2GB DDR2 120gb Sata hd
currently my my.cnf is as follows
Quote:
# This is for a system with little memory (32M - 64M) where MySQL plays # an important part, or systems up to 128M where MySQL is used together with # other programs (such as a web server) # # You can copy this file to # /etc/my.cnf to set global options, # mysql-data-dir/my.cnf to set server-specific options (in this # installation this directory is /var/lib/mysql) or # ~/.my.cnf to set user-specific options. # # In this file, you can use all long options that a program supports. # If you want to know which options a program supports, run the program # with the "--help" option.
# The following options will be passed to all MySQL clients [client] #password = your_password port = 3306 socket = /var/lib/mysql/mysql.sock
according to mysql tuning primer it reports the following
Quote:
-- MYSQL PERFORMANCE TUNING PRIMER -- - By: Matthew Montgomery -
MySQL Version 4.1.20-log x86_64
Uptime = 0 days 17 hrs 52 min 44 sec Avg. qps = 15 Total Questions = 1028682 Threads Connected = 1
Warning: Server has not been running for at least 48hrs. It may not be safe to use these recommendations
To find out more information on how each of these runtime variables effects performance visit: [url] Visit [url] for info about MySQL's Enterprise Monitoring and Advisory Service
SLOW QUERIES Current long_query_time = 5 sec. You have 0 out of 1028694 that take longer than 5 sec. to complete The slow query log is NOT enabled. Your long_query_time seems to be fine
WORKER THREADS Current thread_cache_size = 256 Current threads_cached = 3 Current threads_per_sec = 0 Historic threads_per_sec = 0 Your thread_cache_size is fine
MAX CONNECTIONS Current max_connections = 1000 Current threads_connected = 1 Historic max_used_connections = 4 The number of used connections is 0% of the configured maximum. You are using less than 10% of your configured max_connections. Lowering max_connections could help to avoid an over-allocation of memory See "MEMORY USAGE" section to make sure you are not over-allocating
MEMORY USAGE Max Memory Ever Allocated : 243 M Configured Max Per-thread Buffers : 41 G Configured Max Global Buffers : 75 M Configured Max Memory Limit : 41 G Physical Memory : 1.90 G
Max memory limit exceeds 90% of physical memory
KEY BUFFER Current MyISAM index space = 12 M Current key_buffer_size = 1 M Key cache miss rate is 1 : 14581 Key buffer fill ratio = 44.00 % Your key_buffer_size seems to be too high. Perhaps you can use these resources elsewhere
QUERY CACHE Query cache is enabled Current query_cache_size = 64 M Current query_cache_used = 5 M Current query_cache_limit = 64 M Current Query cache Memory fill ratio = 8.42 % Current query_cache_min_res_unit = 4 K Query Cache is 15 % fragmented Run "FLUSH QUERY CACHE" periodically to defragment the query cache memory If you have many small queries lower 'query_cache_min_res_unit' to reduce fragmentation. Your query_cache_size seems to be too high. Perhaps you can use these resources elsewhere MySQL won't cache query results that are larger than query_cache_limit in size
SORT OPERATIONS Current sort_buffer_size = 32 M Current record/read_rnd_buffer_size = 1020 K Sort buffer seems to be fine
JOINS Current join_buffer_size = 3.00 M You have had 378 queries where a join could not use an index properly You should enable "log-queries-not-using-indexes" Then look for non indexed joins in the slow query log. If you are unable to optimize your queries you may want to increase your join_buffer_size to accommodate larger joins in one pass.
Note! This script will still suggest raising the join_buffer_size when ANY joins not using indexes are found.
OPEN FILES LIMIT Current open_files_limit = 50000 files The open_files_limit should typically be set to at least 2x-3x that of table_cache if you have heavy MyISAM usage. Your open_files_limit value seems to be fine
TABLE CACHE Current table_cache value = 2000 tables You have a total of 908 tables You have 102 open tables. The table_cache value seems to be fine
TEMP TABLES Current max_heap_table_size = 31 M Current tmp_table_size = 30 M Of 3666 temp tables, 0% were created on disk Created disk tmp tables ratio seems fine
TABLE SCANS Current read_buffer_size = 5 M Current table scan ratio = 1544 : 1 read_buffer_size seems to be fine
TABLE LOCKING Current Lock Wait ratio = 1 : 4664 You may benefit from selective use of InnoDB.
I been logging queries for a whole 3 days and my logs is huge and the problem seems to be my phpBB search table thats causing such problems for me heres what I logged
Code: # Time: 070219 19:41:53 # User@Host: admin_db[admin_db] @ localhost [] # Query_time: 11 Lock_time: 0 Rows_sent: 0 Rows_examined: 13274745 SELECT word_id FROM phpbb_search_wordmatch GROUP BY word_id HAVING COUNT(word_id) > 259756;
Everything from phpbb_search_wordmatch takes more then 11 sec Query time.
How can I optimze this and lower the Query time to something like 1 or 2. Some adivce would be helpful me not that good with mysql.
The servers load is vey bad, most of the time 7-8+ and the IO WAIT more than 60% which is decreasing server performace. Please help me to optimzie the MySQL specially. If you have the good my.cnf config please give me the complete config file.
[client] port = 3306 socket = /var/lib/mysql/mysql.sock
anyone have any tips on how to optimize WHM / cpanel to run at top performance on 256-512mb guaranteed VPS accounts?
I've had several VPS providers in the past, and each one of them run webpages and LAMP stuff fine, but as soon as I bring up cpanel /whm, it either crashes the VPS altogether, or cpanel and whm run very slow, choppy, pretty much bogging the whole VPS down, doesn't seem to matter if the VPS is running on 256 or 512, still same issues. I run on Virtuozzo, which I don't know if thats part of the problem, or not,
really all I use out of cpanel/ WHM is the standard things like
creating/deleting accounts RVSkin, Fantastico
basically what a reseller would use. the rest of the stuff I do not use, so I was just wondering if there was a tutorial on how to strip down WHM / cpanel somehow so it would run better on VPS accounts? or if someone could write one up perhaps?
I hope some of you folks could give me some pointers here. I have a server that is used for a gaming clan's purposes. Nothing major as far as games go, but what I got I want running at peak efficiency.
The most important is the Counter-Strike source. I really need the kernel frequency to be tweaked for this. The number one question is can I merely modify the existing config file and recompile that or do I need to start all over? If I could do that is it something like make old or something to that matter?
I am running a couple other servers as well, but there is never a time where all the servers have anyone in them. I keep everything running. After installing some weird stuffs here and there I got the memory usage to go down. I think. Regarding a scenario of having more than one instance of a game, is there any way to share memory between them?
My Swap. Is there a way to force idling games to move to it instead of using physical memory?
The kernel is what I really need to take care of. I'm still somewhat of a n00b and have just switched from centos to debian to top it off. So any tips would make my day. Thanks for reading me.
Hi"Optimization of computing resources has long been an important management issue. One of its aspects concerns server scalability and the question of whether an organization should scale-up or scale out.Assume that the computing performance of the servers can be measured by variable 0 <=p, that their total cost is given by "c" and the relationship between server performance and cost is defined by c=αp^β"
a. What is the cost-performance elasticity(ђ), precisely?
b. What would be the range of values for ђ that would be expected by moore's law and what are its implications?
c. What would be the range of values for ђ that would lead managers to scale-out? Draw a graph and throughly explain the implications.
someone to look at my DB and optimize it as it is running ridiculously high loads of 25. It is vbulletin. I'm only looking for professional companies for this job.
