ipmitool on Solaris 11

[hume.sp...@bofh.ca]

I'm messing around with ipmitool on our new T4 boxes, running Solaris 11.
I'm mostly interested in interrogating the sensors... with hope of bringing
up collectd graphing fan speeds/temperatures/voltages and so on.

ipmitool build from source works, but runs into the occasional timeout on
some sensors. So I figured the ipmitool that comes with the OS would work
better... not quite:

# /usr/sbin/ipmitool -I bmc power status
Error loading interface bmc

The bmc driver is loaded, and /dev/bmc does exist. I don't see any attempt
to access it via truss, though, so "loading" must mean something else. Has
anyone played with this, without using the LAN interface?

--
Brandon Hume - hume -> BOFH.Ca, http://WWW.BOFH.Ca/

[cindy swearingen]

Hi Brandon,

I've seen some funkiness with ipmitool in Solaris 11, fixed in a later
build.
The integrating CRs say that you must use the -I bmc option but you
seem
to have a different issue.

Could there be any conflict between your 2 versions? Did you
completely
uninstall the other version?

What does this output provide:

# modinfo | grep BMC
185 7bbbc000 2da8 13 1 bmc (BMC 1.13)

On my ORCL,SPARC-T3-4, I see this:

# /usr/sbin/ipmitool -I bmc power status

Chassis Power is on

You can see that I need to provide the -I bmc option, here:

# ipmitool sensor
Could not open device at /dev/ipmi0 or /dev/ipmi/0 or /dev/ipmidev/0:
No such file or directory
Get Device ID command failed
Unable to open SDR for reading

# ipmitool -I bmc sensor
/SYS/T_AMB | 13.000 | degrees C | ok | na |
na | na | na | 45.000 | 50.000
MB/V_+1V0 | 1.049 | Volts | ok | 0.931 |
0.941 | 0.960 | 1.137 | 1.156 | 1.166
MB/V_+3V3 | 3.296 | Volts | ok | 2.945 |
3.042 | 3.062 | 3.530 | 3.569 | 3.666
MB/V_+3V3_FMOD | 3.296 | Volts | ok | 2.867 |
2.945 | 3.003 | 3.530 | 3.569 | 3.666
MB/V_+5V0_SBP0 | 4.992 | Volts | ok | 4.563 |
4.758 | 4.875 | 5.226 | 5.304 | 5.577
MB/V_+5V0_SBP1 | 4.992 | Volts | ok | 4.563 |
4.758 | 4.875 | 5.226 | 5.304 | 5.577
MB/V_+12V0 | 12.213 | Volts | ok | 10.797 |
11.387 | 11.505 | 12.508 | 12.626 | 13.216
MB/V_BAT | 3.120 | Volts | ok | na |
na | 2.516 | na | na | na

Thanks, Cindy

[hume.sp...@bofh.ca]

cindy swearingen <cindy.sw...@gmail.com> wrote:
> # modinfo | grep BMC
> 185 7bbbc000 2da8 13 1 bmc (BMC 1.13)

In my case:

# modinfo | grep BMC
183 7bfc8000 2da8 252 1 bmc (BMC 1.13)
# ls -ld /dev/*ipmi*
ls: No match.


The ipmitool that I built, I haven't installed. I've been running it
out of the build tree.

I have a contract, so I can file an SR if I'm really bothered by this.
I was just checking to see if I was running it in some obviously wrong
way.


(Speaking of which: Oracle seems to be getting aggressive about shutting
down self-help sites... you won't get in trouble for checking CRs and the
like on a non-Oracle forum, will you?)

[cindy swearingen]

On Sep 19, 3:39 pm, hume.spamfil...@bofh.ca wrote:

I don't see the same error that you see above "Error loading interface
bmc"
in the S11 ipmitool related CR, which is 7183093, fixed in s11u1_24.
I'm running
a later build but I do have to specify the -I bmc option.

Can you trying running it on another S11 system? That might help rule
out anything on your specific system.

I'll recheck my policy handbook on sharing info with our customers.

Thanks, Cindy

[hume.sp...@bofh.ca]

cindy swearingen <cindy.sw...@gmail.com> wrote:
> I don't see the same error that you see above "Error loading interface
> bmc"
> in the S11 ipmitool related CR, which is 7183093, fixed in s11u1_24.
> I'm running
> a later build but I do have to specify the -I bmc option.

Then maybe/hopefully something's been fixed in U1 that'll make this work.

We're pretty much biding our time until U1 because of a "mondo" queue error
that hangs/crashes the T4s. It's fixed in U1 so I *know* we'll be upgrading
as soon as possible. :)

> Can you trying running it on another S11 system? That might help rule
> out anything on your specific system.

Unfortunately the only two S11 machines we have are two identically
configured T4s. The problem appears on both.

> I'll recheck my policy handbook on sharing info with our customers.

I certainly hope it's okay, but I'd hate to find out later it wasn't.
And... well, it's *Oracle*. :)