Hello all!
I have a strange problem with Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev a0).
I have a few servers with the same cards. All servers with Debian Linux OS.
First server: 2xXeon X5570, 192 GB DDR3 RAM, pluuged to the switch with two SFP-H10GB-CU5M Cisco SFP+ Twinax cables, kernel version: 3.2.0-0.bpo.4-amd64 #1 SMP Debian 3.2.54-2~bpo60+1 x86_64 GNU/Linux, mlx4_en driver verision: 2.3-1.0.0 (Oct 22 2014), hyper trading enabled. Almost all fine (some problems with loading driver), all 20 Gbit/s outgoing traffic generated.
Second server: 2xXeon X5650, 88 Gb DDR3 RAM, pluuged to the switch with one Intel SFP+ SR FTLX8571D3BCV-IT module, kernel version: 3.2.0-4-amd64 #1 SMP Debian 3.2.63-2 x86_64 GNU/Linux, mlx4_en driver verision: 2.3-1.0.0 (Sep 30 2014), hyper trading enabled, all 10 Gbit/s outgoing traffic generated.
Third server: 2xXeon E5645, 256 Gb DDR3 RAM, pluged to the switch with two Intel SFP+ SR FTLX8571D3BCV-IT modules, I tried
both kernels from the first and second server, tried native mlx4_en version: 1.5.4.2 (October 2011) and downloaded 2.3-1.0.0 driver. When hyper trading enabled on both driver version I received something bigger 2 Gbit/s per 2 ports, when I disabled hyper trading, I received all 14 Gbit/s outgoing traffic with mlx4_en driver verision: 2.3-1.0.0, and all 20Gbit/s with native mlx4_en version: 1.5.4.2 but I have too many errors in /var/log/messages
What to do?
