Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
linux:sysctl [2009/03/04 13:31]
greebo
linux:sysctl [2016/02/23 10:52] (current)
zagi
Line 1: Line 1:
 ====== Linux sysctl options ====== ====== Linux sysctl options ======
- 
- 
  
  
Line 63: Line 61:
 # #
 # Do not send ICMP redirects (we are not a router) # Do not send ICMP redirects (we are not a router)
-net/ipv4/conf/all/send_redirects = 1+net/ipv4/conf/all/send_redirects = 0
 # Do not accept IP source route packets (we are not a router) # Do not accept IP source route packets (we are not a router)
-net/ipv4/conf/all/accept_source_route = 1+net/ipv4/conf/all/accept_source_route = 0
  
 # tcp/ip tweak - window size # tcp/ip tweak - window size
Line 88: Line 86:
 net.ipv4.ipfrag_time = 30 net.ipv4.ipfrag_time = 30
  
-net.ipv4.tcp_ecn = 1+net.ipv4.tcp_ecn = 0
 net.ipv4.tcp_syncookies = 1 net.ipv4.tcp_syncookies = 1
 net.ipv4.tcp_timestamps = 1 net.ipv4.tcp_timestamps = 1
 +net.ipv4.tcp_sack = 1
  
 # Linux 2.6 has only 32Mb shared memory # Linux 2.6 has only 32Mb shared memory
Line 99: Line 98:
  
 #net.core.somaxconn=512 #net.core.somaxconn=512
 +
 +# recommended for hosts with jumbo frames enabled
 +net.ipv4.tcp_mtu_probing=1
 +
 +net.ipv4.tcp_allowed_congestion_control = highspeed  lp cubic reno
 +net.ipv4.tcp_congestion_control = highspeed
 +
 +net.ipv4.tcp_slow_start_after_idle=0
 +
 </code> </code>
  
Line 153: Line 161:
  
 ===== Misc add-on options on good to know bases  ===== ===== Misc add-on options on good to know bases  =====
 +
 +==== TCP "thin streams" optimisation in Linux ====
 +
 +If you're using ssh logins over lossy networks (such as many mesh networks), you may be annoyed at the random delays you get after a loss event.  This is due to the fact that modern TCP is optimised for bulk transfer, and that it behaves badly in the presence of packet loss when there are less than 4 packets in flight.
 +
 +Linux 2.6.34 and later is able to use a more aggressive variant of TCP when a given TCP flow is detected as being "thin", i.e. as having less than 4 packets in flight.  While this violates a number of TCP RFCs, the
 +aggressive TCP variant is only used with "thin" streams, and hence should not cause any congestion issues.  However, please do not enable this feature on web servers and more generally systems that handle lots
 +of connections.
 +
 +This optimisation is enabled by putting the following in ''/etc/sysctl.conf'':
 +
 +   net.ipv4.tcp_thin_dupack = 1
 +   net.ipv4.tcp_thin_linear_timeouts = 1
 +
 +Since it's a sender-only modification to TCP, the effect will be most dramatic if you do that on the client.
 +
 +For more information, please see ''/usr/src/linux/Documentation/networking/tcp-thin.txt''
  
 ==== Reboot on kernel panic ==== ==== Reboot on kernel panic ====
linux/sysctl.1236169901.txt.gz ยท Last modified: 2009/05/25 00:34 (external edit)
CC Attribution-Share Alike 4.0 International
Driven by DokuWiki Recent changes RSS feed Valid CSS Valid XHTML 1.0 ipv6 ready