Production of a Linux device concurrency is relatively large, droped packet more, especially in the running game packets, there is a serious loss of packet phenomenon, suspected network card performance, in the replacement of equipment before the solution through the software method, through some data on the internet to show that this phenomenon, Also may be the network card buffer size is too small reason, so try to change the buffer resolution, the following equipment run 64 days, lost more than more than 2 billion packets
Find some foreign articles, you can modify the buffer size of the network card by Ethtool, first of all, to network card support, my server is Intel's 1000M network card, we look at the Ethtool description
-g–show-ringqueries the specified Ethernet device for RX/TX ring parameter information.
-g–set-ringchanges the Rx/tx ring parameters of the specified Ethernet device.
View the current NIC's buffer size ethtool-g eth0
Ring parameters for eth0:
Pre-set maximums:
Rx:4096rx mini:0
RX jumbo:0
tx:4096
Current hardware settings:
rx:256
RX mini:0
RX jumbo:0
tx:256
Since my RX package will have droped, we use ethtool-g eth0 Rx 2048
The same is true for eth1 ethtool-g eth1 Rx 2048
And look at the modified ethtool-g eth0.
Ring parameters for eth0:
Pre-set maximums:
rx:4096
RX mini:0
RX jumbo:0
tx:4096
Current hardware settings:
rx:2048
RX mini:0
RX jumbo:0
tx:2048
A few hours later, there is no increase in packet loss