[openib-general] Other Outstanding Operational Issues with OpenSM 1.1.0

[Hal Rosenstock]

Hi again,

I also see the following operational issues with OpenSM 1.1.0:

With an Anafa 2 based switch, I can see several links keep getting
bounced (port state changes) by OpenSM. This only occurs when OpenSM is
running. As soon as it is killed, this no longer occurs. There are no
significant physical errors that could be trigering the LTSM. Any ideas
on what is going on here ? [This is the most important of these issues.]

In running some SM failover tests, I think there is a minor issue with
the SM state machine. It appears that when a new SM comes up with a
lower GUID and same priority, it takes over from an already established
master. I don't think that is supposed to occur.

Also, still outstanding is the SM Set PortInfo from armed to active
sometimes doesn't work. This was seen in the trace I sent and could also
be seen in Troy's/Brett's osm.log as well. I'm sure we'll see a lot more
of this.

Thanks for your help in chasing these down.

-- Hal