I recently inherited an Infiniband network containing multiple switches, and I know that one of these switches is running the subnet manager. The rest supposedly have that feature turned off, or were never enabled. The trouble is, I have no idea which one it is...
I'd like to replace the switch subnet manager with OpenSM running on a couple of my infrastructure servers. Is there any way, short of logging into each switch individually, to determine which switch is running the SM?
If your infrastructure servers can run OpenSM, they may also have the OFED stack utilities installed as well. If the commands are available, ibstat will tell you the SM's LID. Then use the LID in smpquery ND -L <LID> to get the node description of the node running the SM.
If you have an OFED installed in your system and the package infiniband-diags you can use the command sminfo to list running subnet managers in the Infiniband network:
~ # sminfo
sminfo: sm lid 18 sm guid 0x0000000, activity count 55554951 priority 0 state 3 SMINFO_MASTER
It will list all known subnet managers for that node and the master subnet manager.
First get the current SM manager info with sminfo and then query the node description with the saquery either by LID (local ID on your net), or GUID (global unique ID)
# sminfo
sminfo: sm lid 8 sm guid 0x24be05ffff8a5f71, activity count 53314 priority 14 state 3 SMINFO_MASTER
# saquery -O 8
login2 HCA-1
-OR
# saquery -U 0x24be05ffff8a5f71
login2 HCA-1
In case the node description doesn't say much, one can investigate the InfiniBand net with ibnetdiscover | less -S and look for the GUID and the switch / port it is connected to.
For the list of all the Subnet Managers configured on the fabric, run ibdiagnet and check the ibdiagnet2.sm file (full path should be reported to the standart output).
If you reach your manager by IP address, you could look the ARP/MAC tables of your switches to find which IP address corresponds to your switch.
