NFS over LAGG / lacp poor performance

[Gerrit Kühn]

On Fri, 25 Apr 2014 13:34:06 +0200 Marek Salwerowicz <marek_sal at wp.pl>
wrote about Re: NFS over LAGG / lacp poor performance:



GK> irq256: igb0:que 0              99396134         64
GK> irq257: igb0:que 1              61496018         39
GK> irq258: igb0:que 2             101687742         66
GK> irq259: igb0:que 3             100824264         65
GK> irq260: igb0:link                      2          0
GK> irq261: igb1:que 0               1666960          1
GK> irq262: igb1:que 1            2325576555       1510
GK> irq263: igb1:que 2               1563283          1
GK> irq264: igb1:que 3               1897428          1
GK> irq265: igb1:link                      2          0

MS> For me on storage1 (9.1-RELEASE) it looks like:

MS> irq265: igb0:que 0            2307223482        323
MS> irq266: igb0:link                      4          0
MS> irq267: igb1:que 0             271641638         38
MS> irq268: igb1:link                      6          0
MS> irq269: igb2:que 0              91665104         12
MS> irq270: igb2:link                      6          0
MS> irq271: igb3:que 0             628139928         88
MS> irq272: igb3:link                      5          0

MS> But in my case all igb links are aggregated using LACP (lagg0), then
MS> there are 2 vlans over lagg0 (vlan14 and vlan900) and the vlan900 is
MS> one dedicated for NFS
MS> And I don't have more than one queue per interface

Thanks for your input. As far as I understood so far, there should be one
igb queue created per cpu core in the system by default (and this is what
I see on my system). But my irq rate looks quite high to me (and it is
only on one of these queues).
Maybe I'll try to reduce this to one queue and see what happens. Does
anybody else in here happen to know something about this?


cu
  Gerrit