729323 - Rack panda-001 in Haxxor so Releng can use it in staging environment

Status: RESOLVED FIXED

(Infrastructure & Operations Graveyard :: CIDuty, task, P2)

Description

[cmtalbert]

Hostname: panda01
Serial Number: U4430510373
MAC(s): 26:f9:37:1c:5b:01 (eth0)
IP address(es): none yet
Switch Ports: (the name and port number of any switch you've physically connected the device to)
Management IP: n/a
OOB switch & port: n/a
Status: staging
SystemRack: (where the machine is located: room, row, rack number, and rack unit)
Asset Tag: n/a
Date Purchased: Sept 2011
Operating System: Android 4.0.3
Server Model: Panda Board
Allocated to: A-Team (automation & tools)

Comment 1

[John O'Duinn [:joduinn] (please use "needinfo?" flag)]

per irc w/ctalbert:

(In reply to Clint Talbert ( :ctalbert ) from comment #1)
> Hostname: panda01
we'll call this "panda-001" instead, as we expect to end up with more than 99 of these boards. 


> Serial Number: U4430510373
> MAC(s): 26:f9:37:1c:5b:01 (eth0)
it seems that the board is getting a new mac address on each reboot. ctalbert is investigating.



> IP address(es): none yet
> Switch Ports: (the name and port number of any switch you've physically
> connected the device to)
> Management IP: n/a
> OOB switch & port: n/a
> Status: staging
> SystemRack: (where the machine is located: room, row, rack number, and rack
> unit)
> Asset Tag: n/a
> Date Purchased: Sept 2011
> Operating System: Android 4.0.3
> Server Model: Panda Board
> Allocated to: A-Team (automation & tools)

Comment 2

[cmtalbert]

(In reply to John O'Duinn [:joduinn] from comment #1)
> > MAC(s): 26:f9:37:1c:5b:01 (eth0)
> it seems that the board is getting a new mac address on each reboot.
> ctalbert is investigating.
Yeah, looks like we will need a patched kernel. See: http://blog.bisect.de/2012/01/pandaboard-get-persistent-mac-address.html

Working on a new build with a patched kernel.

Comment 3

[Dustin J. Mitchell [:dustin] (he/him)]

(In reply to Clint Talbert ( :ctalbert ) from comment #0)
> Switch Ports: (the name and port number of any switch you've physically

This will be important as we try to track down network problems - can you fill it in when known?

> Asset Tag: n/a

Let's get an asset tag on there so we can uniquely identify this unit.  The Desktop-support folks should have spare stickers.

Thanks!

Comment 4

[Amy Rich [:arr] [:arich]]

I've added panda-001 to inventory, dns, and dhcp with the specified mac and an IP of 10.250.48.250.  Matt, can you please work with clint to get this board installed into haxxor and fill in the rest of the inventory data after getting the board asset tagged and plugging it in?

https://inventory.mozilla.org/en-US/systems/show/5123/

Comment 5

[cmtalbert]

Yeah, we will fill in everything else once it is in haxxor. We can't put it in haxxor until it has a stable mac address.

Yes, the *mac* address is changed every time the board boots. If I'm understanding the network you guys have in haxxor then that means this system will be useless there until we convince the kernel on the panda board to quit re-generating its mac address.

Comment 6

[Dustin J. Mitchell [:dustin] (he/him)]

Long-term, yes.  In the short term, if it's easier to assign it a static IP even with a changing MAC, that may work.  Of course, ARP timeouts may make it unreachable for some time on every reboot.

Comment 7

[cmtalbert]

(In reply to Dustin J. Mitchell [:dustin] from comment #6)
> Long-term, yes.  In the short term, if it's easier to assign it a static IP
> even with a changing MAC, that may work.  Of course, ARP timeouts may make
> it unreachable for some time on every reboot.

Dustin, I was just about to ask that question.  So if we can assign a static IP, that solve the problem (we can probably do that without re-building the kernel). Can you give me a static IP for this pandaboard?

Comment 8

[Dustin J. Mitchell [:dustin] (he/him)]

It gets this particular board working -- it's not a good long-term solution, as it makes the boards impossible to track down on the network (and we find ourselves doing that sort of thing more often than you'd think).

You can use 10.250.48.250 as Amy assigned in comment 4.

Comment 9

[Matthew Larrain[:MaRu]]

clint can you deliver the pandaboard to my desk on floor 2. I am located by desktop support.

Comment 10

[cmtalbert]

(In reply to Matthew Larrain[:digipengi] from comment #9)
> clint can you deliver the pandaboard to my desk on floor 2. I am located by
> desktop support.

I'll have it for you tomorrow. I put it in fastboot flash mode thinking i was going to rebuild the kernel so I have to blow a new OS on it, and once it's in this mode, I have to be onsite to do it (I'm in SF today).

Comment 11

[Matthew Larrain[:MaRu]]

Device is on my desk I will rack it to a PDU that can control reboots and get it updated in invetory.

Comment 12

[Matthew Larrain[:MaRu]]

I have racked the panda board in Haxxor I will be making the changes needed to inventory and updating the CDU.

Comment 13

[Matthew Larrain[:MaRu]]

CDU info pdu6.df202-1  Branch 1:6

Comment 14

[Armen [:armenzg]]

From my point of view:
* what do I need to do to reach the board?
* what do I need to run any unit tests?

Comment 15

[Joel Maher ( :jmaher ) (UTC -8)]

unit tests won't run unless you manually install fennec via adb. The sutagent needs to be retooled to work with ICS which includes the installation.

We cannot reboot the device without physical intervention right now.

Comment 16

[Armen [:armenzg]]

ctalbert said that I can telnet to his machine (jmaher's board behaves differently and needs adb).

I will look to talk to the sut agent on this board next week.

Comment 17

[Armen [:armenzg]]

The board seems to be PING down but this board won't be helpful if we can't reboot it.

Comment 18

[Matthew Larrain[:MaRu]]

PDU and inventory info updated should be all good to go now.

Comment 19

[Armen [:armenzg]]

All that it could be done is done.
Thanks digipengi.