MOC-R4PAC22U31-S1C failing to boot

[larsks]

I was unable to get node MOC-R4PAC22U31-S1C to boot to an openshift discovery ISO ("No available boot device"). I was also unable to get the virtual console to connect (although the thumbnail screenshot in the idrac web interface was working).

[larsks]

[tzumainn]

Regarding the console - I was able to connect once I ran openstack baremetal node console enable

[tzumainn]

@larsks can I clean the node and try redeploying?

[larsks]

@tzumainn go ahead

[tzumainn]

Okay - I didn't clean the node, but I think I figured out what was going on.

The node has the following ports:

+--------------------------------------+-------------------+
| UUID                                 | Address           |
+--------------------------------------+-------------------+
| 2b18e8be-2650-4e53-b3f4-3580f0534501 | e0:d8:48:e5:ae:a9 |
| 22d5cb31-3481-436f-991e-9c1ffbb3c5ee | e0:d8:48:e5:ae:ac |

The provisioning network was attached to the first port (as is normal); however, the console only shows the node trying to pxe boot using the second MAC address. I attached the provisioning network to the second port as well, and the boot continued on a retry. This is the sort of issue that falls through our tests, because inspection/cleaning/provisioning all attach the provisioning network to all available baremetal ports.

@hakasapl I think this node might have something wrong with the NIC associated with e0:d8:48:e5:ae:a9

[larsks]

@tzumainn what's the right way to mark this node as "broken" so that it doesn't get acquired by anybody?

[tzumainn]

I can handle that - I just wanted to make sure you're not using the node (since it is technically usable).

[larsks]

I have leased this system to the hwbroken project.

[tzumainn]

okay, thanks! the other thing I do (which I just did) is set maintenance to "True" so Ironic doesn't try and control the power state.