Additions:
Moved to Doc Wiki
http://doc.pfsense.org/index.php/Installation_Tuning
http://doc.pfsense.org/index.php/Installation_Tuning
Deletions:
By default we tune the net.inet.ip.intr_queue_maxlen sysctl to 50. Starting with 1.2-RC2 we automatically bump this value to 1500 but this requires a re installation or a manual change to /etc/sysctl.conf.
This will show the current setting:
sysctl net.inet.ip.intr_queue_maxlen
However, in largely loaded installations this may not be enough. Here is how to check.
sysctl net.inet.ip.intr_queue_drops
If the above shows values above 0, try changing net.inet.ip.intr_queue_maxlen *2.
For example: sysctl net.inet.ip.intr_queue_maxlen=3000
Keep performing the above until you find the sweet spot.
Afterwards, edit /etc/sysctl.conf and add the value to it:
IE (without the sysctl )
net.inet.ip.intr_queue_maxlen=3000
=====Wrap tuning and polling=====
//**(As always, make backups of these files before you proceed)**//
If you are running on embedded platforms, start of by issuing the command: /etc/rc.conf_mount_rw
And when you are done editing files, issue the command: /etc/rc.conf_mount_ro
This is only needed for embedded platform !
If you are using the webConfigurator, simply visit Diagnostics, Edit File which does the above for you.
**edit /etc/sysctl.conf to look like this:**
-----------------------------------------------------------
# Do not send RSTs for packets to closed ports
net.inet.tcp.blackhole=2
# Do not send ICMP port unreach messages for closed ports
net.inet.udp.blackhole=1
# Generate random IP_ID's
net.inet.ip.random_id=1
# Breaks RFC1379, but nobody uses it anyway
net.inet.tcp.drop_synfin=1
net.inet.ip.redirect=0
net.inet.tcp.syncookies=1
net.inet.tcp.recvspace=65228
net.inet.tcp.sendspace=65228
net.inet.ip.fastforwarding=1
net.inet.tcp.delayed_ack=0
net.inet.udp.maxdgram=57344
kern.ipc.maxsockbuf=2097152
kern.ipc.somaxconn=1024
kern.polling.burst_max=1000
kern.polling.each_burst=120
kern.polling.idle_poll=1
kern.polling.user_frac=5
kern.polling.reg_frac=50
kern.polling.enable=1
**Edit /boot/loader.conf to look like this:**
------------------------------------
autoboot_delay="3"
loader_color="YES"
mfsroot_load="YES"
mfsroot_type="mfs_root"
mfsroot_name="/boot/mfsroot"
hw.ata.atapi_dma="1"
hw.ata.ata_dma="1"
hint.acpi.0.disabled="1"
hint.apic.0.disabled="1"
kern.hz="532"
//NOTE! If you are on a embedded platform, run /etc/rc.conf_mount_rw to mount the image RW and then run /etc/rc.conf_mount_ro after making the changes or use the Diagnostics -> Edit file feature of the webConfigurator which does this automatically.//
Some of these settings might not be neccessary but i included everything we did to increase performance on the WRAP we tested with.
Tests where done with netio.
------------------------------------
FreeBSDTuning
Additions:
FreeBSDTuning
Additions:
However, in largely loaded installations this may not be enough. Here is how to check.
Deletions:
Additions:
By default we tune the net.inet.ip.intr_queue_maxlen sysctl to 50. Starting with 1.2-RC2 we automatically bump this value to 1500 but this requires a re installation or a manual change to /etc/sysctl.conf.
Deletions:
Additions:
=====Tuning a full installation=====
By default we tune the net.inet.ip.intr_queue_maxlen sysctl to 50. Starting with 1.2-RC2 we automatically bump this value to 1500.
This will show the current setting:
sysctl net.inet.ip.intr_queue_maxlen
However, in large loaded installations this may not be enough. Here is how to check.
sysctl net.inet.ip.intr_queue_drops
If the above shows values above 0, try changing net.inet.ip.intr_queue_maxlen *2.
For example: sysctl net.inet.ip.intr_queue_maxlen=3000
Keep performing the above until you find the sweet spot.
Afterwards, edit /etc/sysctl.conf and add the value to it:
IE (without the sysctl )
net.inet.ip.intr_queue_maxlen=3000
By default we tune the net.inet.ip.intr_queue_maxlen sysctl to 50. Starting with 1.2-RC2 we automatically bump this value to 1500.
This will show the current setting:
sysctl net.inet.ip.intr_queue_maxlen
However, in large loaded installations this may not be enough. Here is how to check.
sysctl net.inet.ip.intr_queue_drops
If the above shows values above 0, try changing net.inet.ip.intr_queue_maxlen *2.
For example: sysctl net.inet.ip.intr_queue_maxlen=3000
Keep performing the above until you find the sweet spot.
Afterwards, edit /etc/sysctl.conf and add the value to it:
IE (without the sysctl )
net.inet.ip.intr_queue_maxlen=3000
