QP INIT to RTR fails with "Connection timed out"

Infrastructure & Networking Software And Drivers Mellanox OFED
1

Hi,

My UC QP makes a transition from INIT to RTR state and throws an error of “Connection timed out”. I have populated the GRH attrs as well. Is there a way to check error logs to investigate the cause of this issue?

Secondly, to investigate the issue, i installed the debug symbols for libibverbs corresponding to the MLNX_OFED kernel i was using and i think i found a small bug in one of the function calls.
I would like to know is there a way to report it?

2

I connected the two ports of my NIC in a loopback connection. The issue was because my IPs and their MAC addresses were not discoverable somehow. Once i ran ib_write_bw command which uses RDMA CM underneath(my application doesn’t use this, instead uses TCP handshake mechanism to communicate QPN and GID), my IPs and their MAC addresses were visible and i could run my application.