netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Ido Schimmel <idosch@idosch.org>
To: Sriram Yagnaraman <sriram.yagnaraman@est.tech>
Cc: "netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	"linux-kselftest@vger.kernel.org"
	<linux-kselftest@vger.kernel.org>,
	"David S . Miller" <davem@davemloft.net>,
	Eric Dumazet <edumazet@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	David Ahern <dsahern@kernel.org>,
	Ido Schimmel <idosch@nvidia.com>, Shuah Khan <shuah@kernel.org>
Subject: Re: [PATCH 3/3] selftests: forwarding: Add test for load-balancing between multiple servers
Date: Tue, 22 Aug 2023 21:57:24 +0300	[thread overview]
Message-ID: <ZOUFFOIQeGazn2Dr@shredder> (raw)
In-Reply-To: <DBBP189MB1433714989BBE41321848336951EA@DBBP189MB1433.EURP189.PROD.OUTLOOK.COM>

On Mon, Aug 21, 2023 at 07:36:47PM +0000, Sriram Yagnaraman wrote:
> Do you think it would be OK to drop this patch from the series for now? I can come back with the selftest when I have something working correctly?

There's a more direct way of testing it and that's by counting the
number of times the relevant FIB trace point was triggered. This script
[1] does it for IPv4. For IPv6 the equivalent trace point is called
fib6:fib6_table_lookup. The script can obviously be made nicer.

Before the patches:

# ./mp_repo.sh 
10020

After the patches:

# ./mp_repo.sh 
65535

You can see that after the patches the trace point is triggered for
every packet. Sometimes it's a bit less. I assume because some events
are lost.

Another approach would be to tweak the current test so that $h1 and $rp1
are configured in a similar fashion to veth0 and veth1.

[1]
#!/bin/bash

ip link del dev veth0 &> /dev/null
ip netns del ns1 &> /dev/null

ip link add name veth0 type veth peer name veth1
ethtool -K veth0 tcp-segmentation-offload off
ethtool -K veth1 generic-receive-offload on
echo 20000 > /sys/class/net/veth1/gro_flush_timeout
echo 1 > /sys/class/net/veth1/napi_defer_hard_irqs

ip netns add ns1

ip link set dev veth0 up
ip address add 192.0.2.1/28 dev veth0

ip link set dev veth1 netns ns1
ip -n ns1 link set dev veth1 up
ip -n ns1 address add 192.0.2.2/28 dev veth1

ip -n ns1 link set dev lo up
ip netns exec ns1 sysctl -w -q net.ipv4.conf.all.forwarding=1
ip netns exec ns1 sysctl -w -q net.ipv4.fib_multipath_hash_policy=1

ip -n ns1 link add name dummy1 up type dummy
ip -n ns1 address add 192.0.2.17/28 dev dummy1
ip -n ns1 neigh add 192.0.2.18 lladdr 00:11:22:33:44:55 nud perm dev dummy1
ip -n ns1 neigh add 192.0.2.19 lladdr 00:aa:bb:cc:dd:ee nud perm dev dummy1
ip -n ns1 route add 198.51.100.0/24 nexthop via 192.0.2.18 nexthop via 192.0.2.19

dmac=$(ip -n ns1 -j link show dev veth1 | jq -r '.[]["address"]')
fout=$(mktemp)
perf stat -o $fout -j -e fib:fib_table_lookup -- \
        mausezahn veth0 -a own -b $dmac -A 192.0.2.1 -B 198.51.100.10 \
        -t udp "sp=12345,dp=0-65535" -q
tail -n 1 $fout | jq '.["counter-value"] | tonumber | floor'

  reply	other threads:[~2023-08-22 18:57 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-08-19 11:48 [PATCH 0/3] Avoid TCP resets when using ECMP for load-balancing between multiple servers Sriram Yagnaraman
2023-08-19 11:48 ` [PATCH 1/3] ipv4: ignore dst hint for multipath routes Sriram Yagnaraman
2023-08-21 11:39   ` Ido Schimmel
2023-08-19 11:48 ` [PATCH 2/3] ipv6: " Sriram Yagnaraman
2023-08-19 11:48 ` [PATCH 3/3] selftests: forwarding: Add test for load-balancing between multiple servers Sriram Yagnaraman
2023-08-21 11:34   ` Ido Schimmel
2023-08-21 19:36     ` Sriram Yagnaraman
2023-08-22 18:57       ` Ido Schimmel [this message]
2023-08-23 12:55         ` Sriram Yagnaraman

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ZOUFFOIQeGazn2Dr@shredder \
    --to=idosch@idosch.org \
    --cc=davem@davemloft.net \
    --cc=dsahern@kernel.org \
    --cc=edumazet@google.com \
    --cc=idosch@nvidia.com \
    --cc=kuba@kernel.org \
    --cc=linux-kselftest@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=shuah@kernel.org \
    --cc=sriram.yagnaraman@est.tech \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).