On Friday, September 10, 2021 7:59:54 PM CEST brian.edmisten@viasat.com wrote:
Simon,
Thanks for responding. We are trying out some different solutions for bonding these radios. For scenarios BATMAN seems really well suited for the problem but we wanted to test this one and see how much work we need to put into it. I saw the same behavior with IV but I'll switch back and check on it. While its up though here is what I am seeing with V.
batctl o [B.A.T.M.A.N. adv 2019.4, MainIF/MAC: eth0/00:0c:29:c5:d2:da (bat0/de:8b:cc:39:d0:69 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] 00:0c:29:53:f8:c9 0.320s ( 10000.0) 00:0c:29:53:f8:dd [ eth2] 00:0c:29:53:f8:c9 0.320s ( 10000.0) 00:0c:29:53:f8:d3 [ eth1] * 00:0c:29:53:f8:c9 0.320s ( 10000.0) 00:0c:29:53:f8:c9 [ eth0]
batctl n [B.A.T.M.A.N. adv 2019.4, MainIF/MAC: eth0/00:0c:29:c5:d2:da (bat0/de:8b:cc:39:d0:69 BATMAN_V)] IF Neighbor last-seen 00:0c:29:53:f8:c9 0.436s ( 10000.0) [ eth0] 00:0c:29:53:f8:d3 0.340s ( 10000.0) [ eth1] 00:0c:29:53:f8:dd 0.116s ( 10000.0) [ eth2]
batctl tg [B.A.T.M.A.N. adv 2019.4, MainIF/MAC: eth0/00:0c:29:c5:d2:da (bat0/de:8b:cc:39:d0:69 BATMAN_V)] Client VID Flags Last ttvn Via ttvn (CRC ) * 33:33:00:00:00:02 -1 [....] ( 1) 00:0c:29:53:f8:c9 ( 2) (0x6b62ac80) * 01:00:5e:00:00:01 -1 [....] ( 2) 00:0c:29:53:f8:c9 ( 2) (0x6b62ac80) * 4e:b3:25:58:bd:15 -1 [....] ( 1) 00:0c:29:53:f8:c9 ( 2) (0x6b62ac80) * 33:33:00:00:00:01 -1 [....] ( 1) 00:0c:29:53:f8:c9 ( 2) (0x6b62ac80)
I do not directly see any of the commands outputting transmit quality I would expect the three ethernet nics to be the same but it is an assumption.
Here is the same info under IV batctl o [B.A.T.M.A.N. adv 2019.4, MainIF/MAC: eth2/00:0c:29:c5:d2:ee (bat0/f2:49:86:e6:ea:aa BATMAN_IV)] Originator last-seen (#/255) Nexthop [outgoingIF] * 00:0c:29:53:f8:d3 0.976s (255) 00:0c:29:53:f8:d3 [ eth1] * 00:0c:29:53:f8:c9 0.944s (251) 00:0c:29:53:f8:c9 [ eth0] * 00:0c:29:53:f8:dd 0.368s (255) 00:0c:29:53:f8:c9 [ eth0] 00:0c:29:53:f8:dd 0.368s (255) 00:0c:29:53:f8:d3 [ eth1] 00:0c:29:53:f8:dd 0.368s (252) 00:0c:29:53:f8:dd [ eth2]
batctl n [B.A.T.M.A.N. adv 2019.4, MainIF/MAC: eth2/00:0c:29:c5:d2:ee (bat0/f2:49:86:e6:ea:aa BATMAN_IV)] IF Neighbor last-seen eth0 00:0c:29:53:f8:c9 0.032s eth1 00:0c:29:53:f8:d3 0.992s eth2 00:0c:29:53:f8:dd 0.384s
batctl tg [B.A.T.M.A.N. adv 2019.4, MainIF/MAC: eth2/00:0c:29:c5:d2:ee (bat0/f2:49:86:e6:ea:aa BATMAN_IV)] Client VID Flags Last ttvn Via ttvn (CRC ) * 33:33:00:00:00:02 -1 [....] ( 2) 00:0c:29:53:f8:dd ( 3) (0x9339b660) * 01:00:5e:00:00:01 -1 [....] ( 3) 00:0c:29:53:f8:dd ( 3) (0x9339b660) * 2a:78:9d:5f:f3:f6 -1 [....] ( 1) 00:0c:29:53:f8:dd ( 3) (0x9339b660) * 33:33:00:00:00:01 -1 [....] ( 2) 00:0c:29:53:f8:dd ( 3) (0x9339b660)
Hi Brian,
thank you very much for providing that output. There is only "TQ" (transmit quality) in BATMAN IV, BATMAN V uses througput based metric instead (in kbit/ s). For Ethernet, it tries to read the Ethernet speed directl, therefore you see those 10000 values.
Anyway, in BATMAN IV the values look close enough (they need to be within 50 TQ points). Just as sanity check, did you enable bonding? It is disabled by default. You can use batctl b 1 to enable it.
Unfortunately there is not really logging code for debugging, so let's try checking the settings. If that doesn't work, I could rebuild and verify ...
Cheers, Simon