Hi Vanias,
One can tell if a path is being used by the counters. From this little snippet:
forwarding: unicast-ip4-chain
[@0]: dpo-load-balance: [proto:ip4 index:26 buckets:16 uRPF:38
to:[366212091:16845756186]]
[0] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1570111449:72225126654]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
[1] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1570111449:72225126654]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
I can see that this prefix was matched by 366212091 packets. 1570111449 went
via 10.10.3.2.
Capture this a few times and you’ll see how the packets are forwarded. You can
turn on adjacency counters if you want that granularity (‘adj counters enable’)
But as I said before, it’s a flow based hash, so if you only have one flow,
you’ll only see counts against one peer. You need many flows to get the load
distribution 1:1:1. In fact the best you’ll get is 6:5:5 (since there are 16
buckets for the 3 choices).
/neale
From: "vnaposto...@gmail.com" <vnaposto...@gmail.com>
Date: Thursday 10 December 2020 at 14:46
To: "Neale Ranns (nranns)" <nra...@cisco.com>
Subject: Re: [vpp-dev] VPP ip route add multiple paths
Hello Neale,
Thanks for the quick answer. What i meant by not working is that traffic is not
going through them once all paths are up. Interfaces are up and running though
I have checked them.
Output with one path:
sudo vppctl -s /run/vpp/cli-vpprouter1.sock sh ip fib 48.0.0.0/8
ipv4-VRF:0, fib_index:0, flow hash:[src dst sport dport proto ] epoch:0
flags:none locks:[adjacency:1, recursive-resolution:1, default-route:1,
nat-hi:2, ]
48.0.0.0/8 fib:0 index:23 locks:2
CLI refs:1 src-flags:added,contributing,active,
path-list:[33] locks:2 flags:shared, uPRF-list:38 len:1 itfs:[2, ]
path:[42] pl-index:33 ip4 weight=1 pref=0 recursive: oper-flags:resolved,
via 10.10.3.2 in fib:0 via-fib:24 via-dpo:[dpo-load-balance:25]
forwarding: unicast-ip4-chain
[@0]: dpo-load-balance: [proto:ip4 index:26 buckets:1 uRPF:38
to:[19927080:916645680]]
[0] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1310268869:60272367974]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
Output with 2nd path added:
sudo vppctl -s /run/vpp/cli-vpprouter1.sock sh ip fib 48.0.0.0/8
ipv4-VRF:0, fib_index:0, flow hash:[src dst sport dport proto ] epoch:0
flags:none locks:[adjacency:1, recursive-resolution:2, default-route:1,
nat-hi:2, ]
48.0.0.0/8 fib:0 index:23 locks:2
CLI refs:1 src-flags:added,contributing,active,
path-list:[34] locks:2 flags:shared, uPRF-list:37 len:2 itfs:[2, 3, ]
path:[40] pl-index:34 ip4 weight=1 pref=0 recursive: oper-flags:resolved,
via 10.10.3.2 in fib:0 via-fib:24 via-dpo:[dpo-load-balance:25]
path:[47] pl-index:34 ip4 weight=1 pref=0 recursive: oper-flags:resolved,
via 10.10.6.2 in fib:0 via-fib:26 via-dpo:[dpo-load-balance:28]
forwarding: unicast-ip4-chain
[@0]: dpo-load-balance: [proto:ip4 index:26 buckets:2 uRPF:37
to:[205189323:9438708858]]
[0] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1487667797:68432718662]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
[1] [@12]: dpo-load-balance: [proto:ip4 index:28 buckets:1 uRPF:36 to:[0:0]
via:[129101773:5938681558]]
[0] [@5]: ipv4 via 10.10.6.2 memif2/0: mtu:9000 next:5
02fe0b17bc3602fe734e64440800
Output with 3rd path added:
sudo vppctl -s /run/vpp/cli-vpprouter1.sock sh ip fib 48.0.0.0/8
ipv4-VRF:0, fib_index:0, flow hash:[src dst sport dport proto ] epoch:0
flags:none locks:[adjacency:1, recursive-resolution:3, default-route:1,
nat-hi:2, ]
48.0.0.0/8 fib:0 index:23 locks:2
CLI refs:1 src-flags:added,contributing,active,
path-list:[33] locks:2 flags:shared, uPRF-list:38 len:3 itfs:[2, 3, 4, ]
path:[42] pl-index:33 ip4 weight=1 pref=0 recursive: oper-flags:resolved,
via 10.10.3.2 in fib:0 via-fib:24 via-dpo:[dpo-load-balance:25]
path:[44] pl-index:33 ip4 weight=1 pref=0 recursive: oper-flags:resolved,
via 10.10.6.2 in fib:0 via-fib:26 via-dpo:[dpo-load-balance:28]
path:[43] pl-index:33 ip4 weight=1 pref=0 recursive: oper-flags:resolved,
via 10.10.12.2 in fib:0 via-fib:27 via-dpo:[dpo-load-balance:29]
forwarding: unicast-ip4-chain
[@0]: dpo-load-balance: [proto:ip4 index:26 buckets:16 uRPF:38
to:[366212091:16845756186]]
[0] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1570111449:72225126654]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
[1] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1570111449:72225126654]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
[2] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1570111449:72225126654]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
[3] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1570111449:72225126654]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
[4] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1570111449:72225126654]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
[5] [@12]: dpo-load-balance: [proto:ip4 index:25 buckets:1 uRPF:28 to:[0:0]
via:[1570111449:72225126654]]
[0] [@5]: ipv4 via 10.10.3.2 memif0/0: mtu:9000 next:3
02fef98b60f802fe107edd530800
[6] [@12]: dpo-load-balance: [proto:ip4 index:28 buckets:1 uRPF:36 to:[0:0]
via:[198665280:9138602880]]
[0] [@5]: ipv4 via 10.10.6.2 memif2/0: mtu:9000 next:5
02fe0b17bc3602fe734e64440800
[7] [@12]: dpo-load-balance: [proto:ip4 index:28 buckets:1 uRPF:36 to:[0:0]
via:[198665280:9138602880]]
[0] [@5]: ipv4 via 10.10.6.2 memif2/0: mtu:9000 next:5
02fe0b17bc3602fe734e64440800
[8] [@12]: dpo-load-balance: [proto:ip4 index:28 buckets:1 uRPF:36 to:[0:0]
via:[198665280:9138602880]]
[0] [@5]: ipv4 via 10.10.6.2 memif2/0: mtu:9000 next:5
02fe0b17bc3602fe734e64440800
[9] [@12]: dpo-load-balance: [proto:ip4 index:28 buckets:1 uRPF:36 to:[0:0]
via:[198665280:9138602880]]
[0] [@5]: ipv4 via 10.10.6.2 memif2/0: mtu:9000 next:5
02fe0b17bc3602fe734e64440800
[10] [@12]: dpo-load-balance: [proto:ip4 index:28 buckets:1 uRPF:36
to:[0:0] via:[198665280:9138602880]]
[0] [@5]: ipv4 via 10.10.6.2 memif2/0: mtu:9000 next:5
02fe0b17bc3602fe734e64440800
[11] [@12]: dpo-load-balance: [proto:ip4 index:29 buckets:1 uRPF:34
to:[0:0] via:[695832443:32008292378]]
[0] [@5]: ipv4 via 10.10.12.2 memif4/0: mtu:9000 next:6
02feec1766b902fe7d3d1e0f0800
[12] [@12]: dpo-load-balance: [proto:ip4 index:29 buckets:1 uRPF:34
to:[0:0] via:[695832443:32008292378]]
[0] [@5]: ipv4 via 10.10.12.2 memif4/0: mtu:9000 next:6
02feec1766b902fe7d3d1e0f0800
[13] [@12]: dpo-load-balance: [proto:ip4 index:29 buckets:1 uRPF:34
to:[0:0] via:[695832443:32008292378]]
[0] [@5]: ipv4 via 10.10.12.2 memif4/0: mtu:9000 next:6
02feec1766b902fe7d3d1e0f0800
[14] [@12]: dpo-load-balance: [proto:ip4 index:29 buckets:1 uRPF:34
to:[0:0] via:[695832443:32008292378]]
[0] [@5]: ipv4 via 10.10.12.2 memif4/0: mtu:9000 next:6
02feec1766b902fe7d3d1e0f0800
[15] [@12]: dpo-load-balance: [proto:ip4 index:29 buckets:1 uRPF:34
to:[0:0] via:[695832443:32008292378]]
[0] [@5]: ipv4 via 10.10.12.2 memif4/0: mtu:9000 next:6
02feec1766b902fe7d3d1e0f0800
Vanias
On 10/12/20 1:03 μ.μ., Neale Ranns (nranns) wrote:
Hello Anonymous,
In order to debug IP forwarding issues I’m going to need more info. Please
collect:
‘sh ip fib <YOUR_PREFIX>’
From a working and non-working configuration.
All FIB load-balancing is per-flow. So if you don’t have enough flows you won’t
[necessarily] get the load distribution that you want.
/neale
From: <vpp-dev@lists.fd.io><mailto:vpp-dev@lists.fd.io> on behalf of
"vnaposto...@gmail.com"<mailto:vnaposto...@gmail.com>
<vnaposto...@gmail.com><mailto:vnaposto...@gmail.com>
Date: Wednesday 9 December 2020 at 16:43
To: "vpp-dev@lists.fd.io"<mailto:vpp-dev@lists.fd.io>
<vpp-dev@lists.fd.io><mailto:vpp-dev@lists.fd.io>
Subject: [vpp-dev] VPP ip route add multiple paths
Hello to everyone,
I am having an issue when trying to apply multiple paths on a running VPP to 3
other VPPs.
Specifically after 2 paths are up when I add the 3rd path one or both the first
two stop working. I am using these commands:
sudo vppctl -s /run/vpp/cli-vpprouter1.sock ip route add 48.0.0.0/8 via
10.10.3.2
sudo vppctl -s /run/vpp/cli-vpprouter1.sock ip route add 48.0.0.0/8 via
10.10.6.2
sudo vppctl -s /run/vpp/cli-vpprouter1.sock ip route add 48.0.0.0/8 via
10.10.12.2
On the same subject I have an issue when trying to set different weights on a 2
route path. As suggested on this
doc https://docs.fd.io/vpp/21.01/de/db4/clicmd_src_vnet_ip.html#clicmd_ip_route
I use:
sudo vppctl -s /run/vpp/cli-vpprouter1.sock ip route add 48.0.0.0/8 via
10.10.3.2 weight 9
sudo vppctl -s /run/vpp/cli-vpprouter1.sock ip route add 48.0.0.0/8 via
10.10.6.2 weight 1
but the traffic on weight 1 route is not the 1/10 of the total traffic, more
like 4/10 of it. And that's the good scenario. When traffic
gets elevated VPP sends most of the traffic on the path with the lower value
rather than what I expected it to do.
All vpps are connected through memif sockets.
Can anyone help me with any of these issues? Is it me I am doing something
extremely wrong?
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#18307): https://lists.fd.io/g/vpp-dev/message/18307
Mute This Topic: https://lists.fd.io/mt/78831750/21656
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-
