-----Original Message----- From: Ido Schimmel idosch@idosch.org Sent: Monday, 21 August 2023 13:35 To: Sriram Yagnaraman sriram.yagnaraman@est.tech Cc: netdev@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
On Sat, Aug 19, 2023 at 01:48:25PM +0200, Sriram Yagnaraman wrote:
Create a topology with 3 hosts, a router each in it's own network namespace. Test IPv4 and IPv6 multipath routing from h1 to h2/h3 via router r1 where a multipath route is setup to load-balance between h2 and h3.
See diagram in the test for more information.
How are you running this test? At least with veth pairs it is passing both before and after the patches. I didn't look into the veth driver, it might not even use the listified path.
I agree, the test is flaky, and it doesn't definitively fail before the patch, nor does it definitively pass after the patch. Checking the packet transmit counters is probably not the best way to test this. I will try to rewrite this selftest using ncat. I use mconnect [0], another test utility to test that a TCP connection succeeds for my own testing, but I guess using that in selftest is not an option.
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?
Also, I'm seeing the following errors during the test:
sysctl: setting key "net.ipv4.fib_multipath_hash_policy": Invalid argument sysctl: setting key "net.ipv6.fib_multipath_hash_policy": Invalid argument