[geeks] Sun SSP monkeying - any SSP 'experts' about?

[JP Hindin]

To circle back on this one, although I haven't pulled out the one
complaining about incomplete initialisation, I have made the other work.

It turns out it takes 10-15 minutes for the SSP to complete negotiation
with the Control Board in the E10k. You fire the SSP up first, power on
the E10k, and then you walk away and have a cup of tea, and when you get
back the fans have spun down somewhat as the SSP has taken control and
told the E10k to chill out. After it makes 'contact' it will assign MAIN
status to the SSP.

I suspect the more domains you have (this machine setup has six
preconfigured) the longer it takes, and presumably if you have a second CB
intalled, it takes even longer than that.

The More You Know.

On the upside, I got my E10k to boot to OBP. I now have to get an OS
installed (the latest release I have a copy of is 4.1.4), and sort through
the system boards to find all the failed components and replace them.

 - JP


On Tue, 2 Sep 2014, Jonathan Katz wrote:
> Anything in the docs I gave you? Also, there is a /var/SUNWsms or
> /var/lib/ssp or something that may have the locks for who's the
> master.
>
> On Tue, Sep 2, 2014 at 6:45 PM, JP Hindin <jplist2008 at kiwigeek.com> wrote:
> > I have ended up with a pair of E10ks (one since dismantled for spares),
> > and a triplet of SSPs. Entertainingly, through surprise and misadventure
> > (and quite possibly related to the eventual decommissioning of both the
> > E10ks), all three of the SSPs are the "Spare" SSPs - not the MAINs.
> >
> > The one that already has the EEPROMs for the unit I'm trying to bring up
> > appears to not be fully installed right - it gives me the following each
> > time I try and run something useful:
> > fo_clt_get_host_role: door_call() failed, errno: 9
> > Unable to determine role of SSP
> > Is SSP initialization complete?
> >
> > The other two (one is a mirror of the other, I suspect one _was_ the MAIN
> > for that E10k, but it's internal disk died and it was simply cloned from
> > the corresponding Spare) do not give me such errors - but won't execute
> > any commands (ignoring the key differences to the E10k) because it's the
> > Spare SSP.
> >
> > All three machines are running SSP v3.5.0. The documentation suggests that
> > the appropriate way to force a fail-over is to run:
> > $ setfailover force
> > However I'm told (by setfailover) that the command can only be run _on_
> > the MAIN... which appears to badly defeat the purpose of having a failover
> > setup, unless it failed in such a graceful way as to allow the Spare to
> > be told to take-over.
> >
> > I've tried doing an ssp-unconfig/ssp-config to rebuild the setup on the
> > first SSP (with the init complaints), but it never actually asks the
> > question "Am I the MAIN SSP?" at the end like the documentation says it
> > should.
> >
> > Frankly, I'm running out of braincells here to come up with something else
> > to try.
> >
> > I'd sure love some suggestions.
> >
> >  - JP
> > _______________________________________________
> > GEEKS:  http://www.sunhelp.org/mailman/listinfo/geeks
>
>
>
> --
> -Jon
> +32 0 486 260 686
> _______________________________________________
> GEEKS:  http://www.sunhelp.org/mailman/listinfo/geeks