[ovs-dev] [PATCH ovn] test: add more tests to IP-buffering unit-test

Thu Jun 4 02:03:54 UTC 2020

Hi Lorenzo,

Thanks for the explanation.
Just need one more clarification, please find the same inline.

Regards,
Ankur

________________________________
From: Lorenzo Bianconi
Sent: Wednesday, June 3, 2020 1:47 AM
To: Ankur Sharma
Cc: ovs-dev at openvswitch.org
Subject: Re: [ovs-dev] [PATCH ovn] test: add more tests to IP-buffering unit-test

On Jun 03, Ankur Sharma wrote:
> Hi Lorenzo,
>
> Please find my comments inline.
>
> Regards,
> Ankur
>
> ________________________________
> From: dev <ovs-dev-bounces at openvswitch.org> on behalf of Ankur Sharma <ankur.sharma at nutanix.com>
> Sent: Monday, June 1, 2020 9:05 PM
> To: Lorenzo Bianconi <lorenzo.bianconi at redhat.com>; ovs-dev at openvswitch.org <ovs-dev at openvswitch.org>
> Subject: Re: [ovs-dev] [PATCH ovn] test: add more tests to IP-buffering unit-test
>
> Hi Lorenzo,
>
> Thanks a lot for the patch.
> I will get back on this by EOD PST tomorrow.
>
> Regards,
> Ankur
> ________________________________
> From: Lorenzo Bianconi <lorenzo.bianconi at redhat.com>
> Sent: Friday, May 29, 2020 7:01 AM
> To: ovs-dev at openvswitch.org <ovs-dev at openvswitch.org>
> Cc: Ankur Sharma <ankur.sharma at nutanix.com>; numans at ovn.org <numans at ovn.org>; dceara at redhat.com <dceara at redhat.com>
> Subject: [PATCH ovn] test: add more tests to IP-buffering unit-test
>
> Check the hv is sending the ICMP request using the FIP as src IP in
> a DVR scenario. Update previous tests introducing source NAT for
> distributed traffic cases
>
> Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi at redhat.com>
> ---
>  tests/ovn.at | 21 ++++++++++++++-------
>  1 file changed, 14 insertions(+), 7 deletions(-)
>
> diff --git a/tests/ovn.at b/tests/ovn.at
> index 15b40ca1e..57f632c4e 100644
> --- a/tests/ovn.at
> +++ b/tests/ovn.at
> @@ -14873,7 +14873,7 @@ ovn_start
>  # Physical network:
>  # Tw0 hypervisors hv[12].
>  # hv1 hosts vif sw0-p0.
> -# hv1 hosts vif sw1-p0.
> +# hv2 hosts vif sw1-p0 and sw0-p1
>
>  send_icmp_packet() {
>      local inport=$1 hv=$2 eth_src=$3 eth_dst=$4 ipv4_src=$5 ipv4_dst=$6 ip_chksum=$7 data=$8
> @@ -14973,6 +14973,9 @@ ovn-nbctl lsp-add sw0 sw0-p1 \

[...]

>
>  # Create FIP on sw0-p0, add a route on logical router pipeline and
>  # ARP request for a unkwon destination is sent using FIP MAC/IP
> +ovn-nbctl lr-nat-del lr0 snat
>  ovn-nbctl lr-nat-add lr0 dnat_and_snat 172.16.1.2 192.168.1.3 sw0-p1 f0:00:00:01:02:04
>  ovn-nbctl lr-route-add lr0 172.16.2.0/24 172.16.1.11
>
>  dst_ip=$(ip_to_hex 172 16 2 10)
>  fip_ip=$(ip_to_hex 172 16 1 2)
>  src_ip=$(ip_to_hex 192 168 1 3)
> -gw_router=$(ip_to_hex 172 16 1 11)
> -send_icmp_packet 2 2 f00000110203 $router_mac0 $src_ip $dst_ip 0000 $data
> -echo $(get_arp_req f00000010204 $fip_ip $gw_router) >> expected
> +gw_router_ip=$(ip_to_hex 172 16 1 11)
> +gw_router_mac=f00000010a0a
> +send_icmp_packet 2 2 f00000110203 $router_mac0 $src_ip $dst_ip 0c1b $data
> [ANKUR]: Confused by above line, ofport 2 on hv2 is sw1-p0, whereas i believe we wanted port to use sw0-p1, i.e ofport3

Hi Ankur,

we are using hv2-vif2 here, so sw0-p1

[ANKUR]:
I am still confused here, yes desired source port should be sw0-p1, i.e hv2-vif2.
However, ofport for same is 3, where as we are using 2 here.

ovs-vsctl -- add-port br-int hv2-vif2 -- \
    set interface hv2-vif2 external-ids:iface-id=sw0-p1 \
    options:tx_pcap=hv2/vif2-tx.pcap \
    options:rxq_pcap=hv2/vif2-rx.pcap \
    ofport-request=3

> +echo $(get_arp_req f00000010204 $fip_ip $gw_router_ip) >> expected
> +send_arp_reply 2 1 $gw_router_mac f00000010204 $gw_router_ip $fip_ip
> +echo "${gw_router_mac}f0000001020408004500001c00004000fe0121b4${fip_ip}${dst_ip}${data}" >> expected
>
>  OVN_CHECK_PACKETS([hv2/vif1-tx.pcap], [expected])
>
> [ANKUR]: May be i am missing something, but how are we validating buffering here. We should be validating the pcap on some external endpoint right,
> For example, generate a icmp packet from internal port to an external endpoint, simulate an ARP reply and then look at tx pcap on destination.

this is what I am currently doing since:
- we are sending an ICMP request from an internal port(sw0-p1 -- hv2-vif2)
  for the external destination 172.16.2.10
- the logical router buffers the outgoing ICMP request and generates
  an ARP request using FIP IP/mac as src addresses (172.16.1.2/f0:00:00:01:02:04)
  for the external gw (172.16.1.11)
- we simulate an ARP reply from 172.16.1.11
- we checks in hv2/vif1-tx.pcap (hv2-vif1 is associated with the external
  interface) we have:
  - ARP request
  - ARP reply
  - ICMP request with L2 destination address resolved

is it what you are referring to?

Regards,
Lorenzo

[ANKUR] Got it now, we are simulating sw1-p0 as the gateway. Makes sense, this part is clear to me.

>
>
> --
> 2.26.2
>
> _______________________________________________
> dev mailing list
> dev at openvswitch.org
> https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.openvswitch.org_mailman_listinfo_ovs-2Ddev&d=DwICAg&c=s883GpUCOChKOHiocYtGcg&r=mZwX9gFQgeJHzTg-68aCJgsODyUEVsHGFOfL90J6MJY&m=WlGoK4pEkybzFr5FvyLYn_x3hZuLSfQcHJVgmPbU548&s=UJvQpPo9180SVm6LdvHiFtsdKo6eGyatkCN8NxS4z4I&e=