ipmi chassis power on does not work

Stefan P spommere at gmail.com
Wed Feb 18 16:37:08 CST 2009


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Here's what happened:

When the server is powered down (this is true of many DELL servers
with BMC's), the NIC is put into a "power saving mode," which limits
the NIC to 100Mb/sec.  This would not be a problem except for the
fact that some of our switch ports are configured to run only at
1000Mb/sec, so they will not auto-negotiate down to the NIC's lowered
speed.  Consequently, after the IPMI "power off", the NIC goes to
100Mb, the switch doesn't and the link between the switch and the BMC
is useless until the server is powered up again via DRAC, and the NIC
auto-negotiates back to 1Gb.




-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (MingW32)

iD8DBQFJnI1+Jpvk3d0dmrARAqCrAKC11V7SAS64AyIaE4m3mOPchjA4zgCeJ3et
q6vvS6yxV3PnoEN+rpDFC6E=
=EQIG
-----END PGP SIGNATURE-----
On Tue, Feb 10, 2009 at 9:41 AM, Alexander Dupuy <alex.dupuy at mac.com> wrote:

> Stefan Pommere(?) writes:
>
>> we have a series of PowerEdge 2850, BIOS version A04, BMC version
>> 1.07. They run the same Linux kernel version, and IPMI is configured
>> to work on these machines.
>>
>> Most machines exhibit the problem that a 'power on' and 'power status'
>> after a 'power off' is not possible.
>>
>> # /usr/bin/ipmitool -H staiw04-vip -U user -P pwd chassis power off
>> Chassis Power Control: Down/Off
>> ...
>> ...
>> # /usr/bin/ipmitool -H staiw04-vip -U user -P pwd chassis power status
>> Error: Unable to establish LAN session
>> ipmi_lan_send_cmd failed to open intf
>> Unable to get Chassis Power Status
>>
>> Only a few machines don't show this problem, and between a 'power off'
>> and 'power on', all 'power status' commands succeed.
>>
>> It appears that the 'power off' also cuts power to the BMC;
>> interestingly, access to the DRAC of the problematic node is fine.
>>
>> We'd appreciate any information on what could be causing this problem,
>> how to resolve it, or at least what other diagnostics are required to
>> identify the problem.
>>
>>
>
> I'm not sure that this will help, but it is worth considering that perhaps
> the power has not been cut to the BMC, but rather that the network interface
> (which is shared with the host, unlike the DRAC, which has a dedicated
> network interface) has been disabled.  I know that with a running Linux
> system (PE1950), if you ifconfig down eth0, the BMC will not be able to
> access the network, even though it is still running (and may be accessible
> via the serial port, or via the local kernel IPMI driver).
>
> It seems to me that it is possible, depending on the various BMC and BIOS
> configuration settings, that the chassis power off command is starting an
> operating system soft shutdown (this is the part I'm least sure about - it
> may be that the BMC power off functionality is always a hard power cut) and
> that on some of your systems, the shutdown procedures are explicitly
> deconfiguring the network interfaces, disconnecting the BMC from the
> network, and making it seem like there has been a power cut.  If you have a
> physical serial interface that you can connect (and set the BIOS and BMC
> settings so that you can access the BMC via that serial port) you could see
> if there is any response on the serial port of the system after a chassis
> power off.  (I don't know if the DRAC would provide any such capability
> through its own serial interface, if any).
>
> If the BMC is indeed still powered, but just disconnected, you could
> compare the system shutdown scripts and BIOS power button settings on the
> different machines to understand what might be causing the different
> behavior.
>
> @alex
>
> --
> mailto:alex.dupuy at mac.com
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.us.dell.com/pipermail/linux-poweredge/attachments/20090218/0105382c/attachment.htm 


More information about the Linux-PowerEdge mailing list