[prev in list] [next in list] [prev in thread] [next in thread] 

List:       quagga-dev
Subject:    [quagga-dev 11280] Re: [PATCH] bgpd: bgpd-fix-socket-scale.patch
From:       David Lamparter <equinox () opensourcerouting ! org>
Date:       2014-05-15 18:36:46
Message-ID: 20140515183646.GF602738 () jupiter ! n2 ! diac24 ! net
[Download RAW message or body]

When your patch description has an enumerated list of changes like this:

On Wed, Jan 08, 2014 at 05:31:22PM -0800, Vipin Kumar wrote:
> 1.
> BGP was setting sockets to be non-blocking only for the accepted passive peers.
> As a fix, setting the BGP sockets to be non-blocking even for the active peers.
> 
> 2.
> Keeping bgp_notify_write as non-blocking. It should help BGP's convergence
> in the scenarios when peers are flapping. We believe keeping BGP notifications
> blocking creates more issues than it solves.
> 
> 3.
> BGP does handle the cases of TCP socket buffer full by queuing a write event
> back, there is no functional issue there as such. Still, increasing the peer
> socket buffer size should help reduce event queueing in BGP.

... then it should really be 3 patches.  This matters not only for
reviewability, but also because item 2. of your patch:

diff --git a/bgpd/bgp_packet.c b/bgpd/bgp_packet.c
> -  /* Put socket in blocking mode. */
> -  val = fcntl (peer->fd, F_GETFL, 0);
> -  fcntl (peer->fd, F_SETFL, val & ~O_NONBLOCK);
> -

... was already fixed in 8ff202e "bgpd: write NOTIFY non-blockingly",
dated July 31st 2013.

I've split off item 1. and 3. and applied.  Thanks.

-David

_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
https://lists.quagga.net/mailman/listinfo/quagga-dev
[prev in list] [next in list] [prev in thread] [next in thread]