| View previous topic :: View next topic |
| Author |
Message |
sherayusuf3
Joined: 18 Jan 2009
Posts: 88
Location: Jakarta Indonesia
|
 Posted: Nov 16, 2009 11:02 am Post subject: |
 |
|
| cuty wrote: | There are benchmarks out there with that module.You dont have to take my word for it.
:Note..the more you increase max connections is the more resources will be be sucked up, so have a beast server.
Using nginx produces a considerable difference in cpu usage,but I still kinda prefer some servers to use apache, because when the cpu load get fairly( i.e 30 points )high using nginx , or there is a DDOS attack at the time causing it, nginx times out or takes up to 3mins to start the download.Where as apache the load has to be about 900 for you to notice. |
im already test Xfilesharing work perfect on litespeed web server, but is not free
maybe u can use it
oke try tweak u system kernel
for big size HD add on /etc/rc.local
| Code: | echo 1024 > /sys/block/sda/queue/read_ahead_kb
echo 256 > /sys/block/sda/queue/nr_requests |
for + 512 RAM add on /etc/sysctl.conf
| Code: | vm.swappiness=50
kernel.sem = 250 32000 100 128
kernel.shmall = 2097152
#A single application can allocate up to 2 MB
kernel.shmmax = 2147483648
# minimal allocation unit
kernel.shmmni = 4096
fs.file-max = 256000
vm.vfs_cache_pressure = 50
net.core.rmem_default = 16777216
net.core.rmem_max = 16777216
net.core.wmem_default = 16777216
net.core.wmem_max = 16777216
net.ipv4.tcp_rmem = 10240 87380 16777216
net.ipv4.tcp_wmem = 10240 87380 16777216
net.ipv4.tcp_no_metrics_save = 1
net.ipv4.tcp_window_scaling = 1
net.ipv4.tcp_timestamps = 1
net.ipv4.tcp_sack = 1
net.core.netdev_max_backlog = 5000 |
please post u output
| Code: | cat /proc/sys/vm/overcommit_memory
cat /proc/sys/vm/overcommit_ratio |
|
|
| Back to top |
|
 |
PilgrimX182
Joined: 22 Mar 2006
Posts: 2109
Location: UFO Lab
|
| Posted: Nov 16, 2009 5:42 pm Post subject: |
|
|
| sherayusuf3, my personal thanks for sharing this. |
|
| Back to top |
|
|
cuty
Joined: 14 Apr 2009
Posts: 105
|
| Posted: Nov 16, 2009 9:16 pm Post subject: |
|
|
| sherayusuf3 wrote: | | cuty wrote: | There are benchmarks out there with that module.You dont have to take my word for it.
:Note..the more you increase max connections is the more resources will be be sucked up, so have a beast server.
Using nginx produces a considerable difference in cpu usage,but I still kinda prefer some servers to use apache, because when the cpu load get fairly( i.e 30 points )high using nginx , or there is a DDOS attack at the time causing it, nginx times out or takes up to 3mins to start the download.Where as apache the load has to be about 900 for you to notice. |
im already test Xfilesharing work perfect on litespeed web server, but is not free
maybe u can use it
oke try tweak u system kernel
for big size HD add on /etc/rc.local
| Code: | echo 1024 > /sys/block/sda/queue/read_ahead_kb
echo 256 > /sys/block/sda/queue/nr_requests |
for + 512 RAM add on /etc/sysctl.conf
| Code: | vm.swappiness=50
kernel.sem = 250 32000 100 128
kernel.shmall = 2097152
#A single application can allocate up to 2 MB
kernel.shmmax = 2147483648
# minimal allocation unit
kernel.shmmni = 4096
fs.file-max = 256000
vm.vfs_cache_pressure = 50
net.core.rmem_default = 16777216
net.core.rmem_max = 16777216
net.core.wmem_default = 16777216
net.core.wmem_max = 16777216
net.ipv4.tcp_rmem = 10240 87380 16777216
net.ipv4.tcp_wmem = 10240 87380 16777216
net.ipv4.tcp_no_metrics_save = 1
net.ipv4.tcp_window_scaling = 1
net.ipv4.tcp_timestamps = 1
net.ipv4.tcp_sack = 1
net.core.netdev_max_backlog = 5000 |
please post u output
| Code: | cat /proc/sys/vm/overcommit_memory
cat /proc/sys/vm/overcommit_ratio |
|
why would I do this? I'm not having any issues. |
|
| Back to top |
|
|
sherayusuf3
Joined: 18 Jan 2009
Posts: 88
Location: Jakarta Indonesia
|
| Posted: Nov 17, 2009 1:48 am Post subject: |
|
|
| Quote: | | sherayusuf3, my personal thanks for sharing this. |
Wlcm Pilgrim182
| Quote: | | why would I do this? I'm not having any issues. |
let me explain
| Code: | #A single application can allocate up to 2 MB
kernel.shmmax = 2147483648 |
| Code: | # minimal allocation unit
kernel.shmmni = 4096 |
Kernel parameter SHMMAX by default is too small on Linux
As a consequence of this, an Informix instance that needs to allocate one shared memory segment of 320Mb , will actually use 10 Linux segments of 32Mb. This issue causes a performance problem due to excessive operation system shared memory allocation.
This new kernel setting will be lost if the machine reboots. To change it so that it takes effect every time the system reboots, edit the file /etc/sysctl.conf and add the following line:
why u apache consume higt Memory? can u explain?
allocated Phisycal ram:Swap area 50:50
If you have on average a lot of memory free (or used for cache) you probably want to prevent your system to start swapping. This is done by setting vm.swappiness to 0. In my personal experiences this is what you want for ubuntu with 512 mb of memory or more
| Code: | echo 1024 > /sys/block/sda/queue/read_ahead_kb
echo 256 > /sys/block/sda/queue/nr_requests |
As Linux comes out of the box, the Linux kernel does not expect to work with a single large disk of several TByte in size. A few tuning parameters come to rescue. Note though, the effect of these settings is highly dependent on the workload of your machine. Larger numbers are not necessarily better.
First increase the read-a-head of your RAID device. The number is given in kilo Bytes. The default is 128 KByte. We set it to 1 MByte in this example.
The 2.6.x Linux kernel has seen quite some work put into optimizing disk access by properly scheduling the IO requests. It seems that the queue depth of the device and the scheduler interact somehow. I have not looked at the code, but mailing list evidence suggests that things work better if the device queue depth is lower than the scheduler depth, so:
you can use it if you want, but I'm not forcing you to use it
| Quote: | | why would I do this? I'm not having any issues. |
my bad, sorry it hink u is Thread Started |
|
| Back to top |
|
|
cuty
Joined: 14 Apr 2009
Posts: 105
|
| Posted: Nov 17, 2009 2:07 am Post subject: |
|
|
I have no doubt that the info you copied here is very useful .
thanks for sharing |
|
| Back to top |
|
|
ihavenoidea
Joined: 14 Oct 2009
Posts: 19
|
| Posted: Dec 06, 2009 7:55 pm Post subject: |
|
|
Has anyone tried using SpeedyCGI?
The highest process is the index.cgi on my main server which is literally KILLING my server (DUAL XEON 3GHZ, 8GB RAM). |
|
| Back to top |
|
|
chennaihomie
Joined: 19 Dec 2009
Posts: 8
Location: India
|
| Posted: Dec 19, 2009 6:14 pm Post subject: |
|
|
There is fairly lot of tuning that can be done to server. Some basic things are
1) remove all un-necessary apache module.
2) use hosted DNS with your domain registrar. This can offload some amount of load on the server.
3) stop all un-necessary services on server.
4) increase max connections in apache. This can increase your number of connection but will increase your memory usage.
5) Tunneling static files with lighttpd or ngix can offload a lot of load. |
|
| Back to top |
|
|
komi
Joined: 27 Nov 2009
Posts: 159
|
| Posted: Dec 19, 2009 6:56 pm Post subject: |
|
|
| chennaihomie wrote: | | 5) Tunneling static files with lighttpd or ngix can offload a lot of load. |
Hi, can you tell us how this can be done with a working XFS? I'm interested in doing this. |
|
| Back to top |
|
|
chennaihomie
Joined: 19 Dec 2009
Posts: 8
Location: India
|
|
| Back to top |
|
|
komi
Joined: 27 Nov 2009
Posts: 159
|
| Posted: Dec 19, 2009 10:20 pm Post subject: |
|
|
| Thanks. Did you already manage to do this on your own XFS site? |
|
| Back to top |
|
|
chennaihomie
Joined: 19 Dec 2009
Posts: 8
Location: India
|
| Posted: Dec 19, 2009 10:24 pm Post subject: |
|
|
| no, my site gets only about 2k visits per day (max has gone upto 6k) and it easily manages on apache, Might implement this in future. I have seen sites implementing this for static content delivery (image hosting sites). |
|
| Back to top |
|
|
komi
Joined: 27 Nov 2009
Posts: 159
|
| Posted: Dec 19, 2009 11:40 pm Post subject: |
|
|
| I understand. Thanks for the info. |
|
| Back to top |
|
|
sherayusuf3
Joined: 18 Jan 2009
Posts: 88
Location: Jakarta Indonesia
|
| Posted: Sep 01, 2010 2:11 pm Post subject: |
|
|
Up Up Up
Sundul Gan!!! |
|
| Back to top |
|
|
|
Register
Search
Profile
Log in to check your private messages
Log in
