Re: [Wg-test-framework] Baroque 0 & 1 COLO labeling and cabeling

[Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx>]

Paul L. George writes ("Baroque 0 & 1 COLO labeling and cabeling"):
> I went to the COLO and checked labels on the servers Baroque 0 & 1.
> They were correctly labeled.

On both front and back ?

> Baroque 0 was plugged into PDU-0 4, Comtrol-1-6,  48 port 7  - These are all
> Physical Addresses
> 
> Baroque 1 was plugged into PDU-1-4, Comtrol-1-7, 48 port 8  - These are all
> Physical Addresses

I have checked the osstest control data.  The PDU and serial ports
correspond to my records.

osstest does not need the switch port, so I can't confirm what was
beforehand.

> I have included the Chart I am working from.  Please see if it is
> matches your cable map.

That spreadsheet is equivalent to[1] the version from which I
generated the osstest control data.

[1] For some reason the spreadsheet is not byte-for-byte identical to
the one we have in colo-for-osstest.git, but I assume that's just
something that Microsoft spreadsheet software does.  To verify its
equivalence I loaded it into LibreOffice and exported it as CSV.

> According to this information I am at a loss as to how I could have
> swapped out the wrong motherboard in Baroque 1

The spreadsheet contains the MAC addresses of both machines' NICs.
But you don't say that you checked the MAC addresses.

Given the physical machine, the MAC address can be found in a variety
of ways:

 * Boot into the BIOS.  There is a page in the BIOS setup which gives
   the NIC1 MAC address.

   It is this information source - collected via the serial port -
   which I used to collect the details reported in my mail earlier
   this week, and which led me to conclude that the wrong machine's
   motherboard had been replaced.

 * Boot into an operating system and ask it.  In the case of Linux,
   the MAC address is reported in boot messages (which are captured by
   the osstest serial logging) and can also be printed out using
   ifconfig and other tools.

   It is the boot messages from Linux - collected by osstest's serial
   logging - which provide for me confirmation that the information in
   osstest's control data corresponds to the previous situation.

Note that from my point of view both of these sources of information
came via the serial port, so an error in the serial port plumbing
would mislead me.

Perhaps both the PDU and the serial port wiring was incorrect
previously (that is, previously, the wiring was precisely swapped),
but this was corrected as a side effect of your removal and
replacement of the machines.


All that being as it may be, I am certain that the problem motherboard
was the one which osstest previously regarded as baroque1, that is the
motherboard whose NIC has address 00:1e:67:99:46:e4.

I know this because it is the NIC address which osstest primarily uses
to identify the machine.


> The only sure way I can think to easily see if we are on the same
> page is to turn off Baroque 1 and I will check it is the same system
> physically that you think it is logically.

Checking on the VGA console would allow us to confirm the NIC MAC
address.  If you do this in the BIOS and leave the BIOS in the
relevant, screen I think the serial redirection will allow me to peer
over your shoulder - and if not I can double-check it anyway.

But that will have to wait for Monday now, given our moratorium on
Friday data centre work.


Thanks,
Ian.

_______________________________________________
Wg-test-framework mailing list
Wg-test-framework@xxxxxxxxxxxxxxxxxxxx
http://lists.xenproject.org/cgi-bin/mailman/listinfo/wg-test-framework