[ClusterLabs] Q: "(bnxt_en): transmit queue 2 timed out"

[Marek Marcola]

Hello,

 > Oct 05 20:13:25 h19 kernel: NETDEV WATCHDOG: p4p1 (bnxt_en): transmit queue 2 timed out
I have seen this error few times on HP DL360 with centos7 and on HP ML350 with esxi6.

Disabling offloading on network card on centos7 seems to solve problem for me:
  - RX/TX Checksumming:
     # ethtool -K eno1 rx off tx off
  - TCP Segmentation Offload:
     # ethtool -K eno1 tso off
  - Generic Segmentation Offload:
     # ethtool -K eno1 gso off
  - Scatter and Gather:
     # ethtool -K eno1 sg off

Maybe too much, but this error triggered pacemaker node fencing
and there was no time for deep troubleshooting.

To check current offload status use:
    # ethtool -k eno1

Best regards,
Marek Marcola <Marek.Marcola at gmail.com>