netdev
[Top] [All Lists]

Re: V2.4 policy router operates faster/better than V2.6

To: jeremy.guthrie@xxxxxxxxxx
Subject: Re: V2.4 policy router operates faster/better than V2.6
From: Robert Olsson <Robert.Olsson@xxxxxxxxxxx>
Date: Thu, 13 Jan 2005 09:56:19 +0100
Cc: netdev@xxxxxxxxxxx, Robert Olsson <Robert.Olsson@xxxxxxxxxxx>
In-reply-to: <200501121723.21969.jeremy.guthrie@xxxxxxxxxx>
References: <Pine.LNX.4.44.0501071416060.5818-100000@xxxxxxxxxxxxxxxxxxxxx> <200501121647.30934.jeremy.guthrie@xxxxxxxxxx> <16869.45158.194375.398326@xxxxxxxxxxxx> <200501121723.21969.jeremy.guthrie@xxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
Jeremy M. Guthrie writes:

 > >  As your traffic looks sane double your bucket size of the route hash to
 > >  start with. Use the boot option w. rhash_entries. look at rtstat

 > Does it make sense that the driver would kill throughput and force us to 
 > output these types of messages?

 No the other way around... Higher (driver/cpu) throughput/load causes more 
 dst-entries to freed and you reach the max_size*ip_rt_gc_min_interval which
 is a constant and get "dst cache overflow".

 Increasing  rhash_entries is the easieast way to attack this.. Give this
 a try. Monitor with rtstat. You might even have to quadruple your size.

                                                --ro

<Prev in Thread] Current Thread [Next in Thread>