[<prev] [next>] [day] [month] [year] [list]
Message-ID: <CAONBAafVP9SmprVK3f0SSz_FaMm6YH0wxMjmLJE04HvoKSM2vQ@mail.gmail.com>
Date: Fri, 27 Feb 2015 17:01:57 +0100
From: Ronald Pina <pinaronald@...il.com>
To: netdev <netdev@...r.kernel.org>
Subject: Change network device scheduling weight with sysfs using different
wights for different devices.
Hello everyone
I am trying to change network device scheduling weight with sysfs, the
basic idea is that if i i have 3 network interfaces and i would like
to give more priority to interface eth0 and smaller priority on eth1
and eth2. On the ingress traffic giving eth0 a weight of 64 means
that at every visit of the poll function will dequeue 64 frames, and
since the net_dev are are scheduled in round-robin manner on
net_rx_action , i would like to use a weight of 32 for eth1 meaning
low priority and a smaller time to dequeue frames the same for eth2
too. I am trying to achieve that ,it seems that the file for
configuring the weight is read only. Any help ?
Another issue i was thinking, in order to reduce the packet drop rate
of eth0 in case of a congestion , maybe would be experimental solution
to change the value RAND_LIE variable ( since this variable is a
dependent of mean of frames entering the queue and and_max_backlog,
maybe we can manipulate this variable to minimize the packet drop of
eth0 ).
What do you think , how valuable is this idea?
I have try achieving that using traffic control tools on the egress
queue , the results are great, but i want to put hands on kernel
development and hacking code to achieve near results but this time on
the receive side on net_rx_action.
Best Regards
Ronald
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists