'osm_sa_mad_ctrl_unbind: ERR 1A11: No previous bind' error messages are logged suddenly

Below error messages are logged suddenly.

  • opensm.log

osm_sm_vendor_bind: ERR 5426: Unable to register class 129 version 1

osm_sm_mad_ctrl_bind: ERR 3118: Vendor specific bind failed

osm_sm_bind: ERR 2E10: SM MAD Controller bind failed (IB_ERROR)

perfmgr_mad_unbind: ERR 5405: No previous bind

osm_congestion_control_shutdown: ERR C108: No previous bind

osm_sa_mad_ctrl_unbind: ERR 1A11: No previous bind

I searched about these messages and found that messages are logged when starting opensm service.

In my case, opensm service was already running and there is no action for stop/start opensm service.

So, I do not know the reason of these messages.

I am wondering if you can share about any information for my case.

Thank you.​

OFED version is

GUID in opensm.conf and HCA adapter are same.

Hello Jae,

Thank you for posting your inquiry on the NVIDIA Networking Community.

Based on the information provided, it seems an entity (user or automated process) tried to start another OpenSM instance on the same node, or start the instance, while the original instance was not fully stopped/killed.

Please check the audit logs from the node.

Thank you and regards,

~NVIDIA Networking Technical Support