NFS over LAGG / lacp poor performance
Marek Salwerowicz
marek_sal at wp.pl
Fri Apr 25 11:40:56 UTC 2014
Hello Gerrit,
W dniu 2014-04-25 11:37, Gerrit Kühn pisze:
> On Fri, 25 Apr 2014 09:48:36 +0200 Marek Salwerowicz <marek_sal at wp.pl>
> wrote about NFS over LAGG / lacp poor performance:
>
>
>
> Could you have a look on your system and let me know how your interrupts
> are spread?
For me on storage1 (9.1-RELEASE) it looks like:
> storage1% vmstat
> -i
>
> interrupt total rate
> irq1: atkbd0 77 0
> irq16: ehci0 11355459 1
> irq23: ehci1 26468060 3
> cpu0:timer 565127221 79
> irq264: mps0 1905530055 267
> irq265: igb0:que 0 2307223482 323
> irq266: igb0:link 4 0
> irq267: igb1:que 0 271641638 38
> irq268: igb1:link 6 0
> irq269: igb2:que 0 91665104 12
> irq270: igb2:link 6 0
> irq271: igb3:que 0 628139928 88
> irq272: igb3:link 5 0
> irq273: ahci0 1579878 0
> cpu1:timer 283555294 39
> cpu4:timer 285185117 40
> cpu10:timer 305482789 42
> cpu3:timer 239294364 33
> cpu7:timer 406658435 57
> cpu2:timer 353421522 49
> cpu8:timer 437284694 61
> cpu11:timer 261929365 36
> cpu5:timer 282975629 39
> cpu6:timer 248355888 34
> cpu9:timer 256679915 36
> Total 9169553935 1287
> marek at storage1:/home/marek%
But in my case all igb links are aggregated using LACP (lagg0), then
there are 2 vlans over lagg0 (vlan14 and vlan900) and the vlan900 is one
dedicated for NFS
And I don't have more than one queue per interface
Cheers,
Marek
--
Marek Salwerowicz
More information about the freebsd-net
mailing list