Red Hat Bugzilla – Bug 1302207
introspection shows timeout
Last modified: 2016-04-18 03:02:27 EDT
Description of problem:
When we trying to inspect our overcloud node (Lenovo RD640). the inspection failed with timeout. From our tcpdump, the Bootstrap is working well and get undionly.kpxe , but we don't know why the server stop requesting other files, like kernel and ramdisk. Could you have a look? Thanks.
Attach the tcpdump before the director and node.
[stack@director ~]$ openstack baremetal introspection status 8bdade83-a9bf-44e6-b6ea-4813a1cf6d9b
| Field | Value |
| error | Introspection timeout |
| finished | True |
Version-Release number of selected component (if applicable):
openstack baremetal introspection bulk start
Steps to Reproduce:
The PXE boot failed.
Boot from PXE successfully.
Created attachment 1118702 [details]
Hi! I suspect iPXE firmware problem that some people have reported, but it's hard to tell precisely. Could you please replace file in /httpboot with http://boot.ipxe.org/undionly.kpxe and restart introspection to see if it solves the problem? Otherwise, could you check if this hardware supports any means to look at its virtual console during boot?
Thanks for your checking. I found I don't have the kpxe file in my directory.
[root@director httpboot]# ll
-rwxr-xr-x. 1 root root 5154912 Feb 14 21:33 agent.kernel
-rw-r--r--. 1 root root 340459405 Feb 14 21:34 agent.ramdisk
-rw-r--r--. 1 ironic ironic 260 Feb 14 05:34 inspector.ipxe
So I just copy the kpxe file to there and do introspection again but no lucky.
The hardware is very new and we have also seen the PXE boot phase also (even failed).
And I have tried to use a common virtual server trying to boot from this pxe server. It seems working. So I believe you are right this issue is may at the hardware server side. I will do more checking on it and get back to you.
My bad, the directory with the kpxe file is /tftpboot, not /httpboot. Could you try putting the downloaded file there?
Yes, I realized that also and put the files into /tftpboot, but still cannot boot. As I said later, it seems working within a common virtual server. So I am focusing the physical server iPXE configuration. I will get back to you later.
It might happen that the server has its own iPXE firmware, then replacing files in /tftpboot won't affect it. The vendor might provide an update in this case.
This issue is fixed by changing the PXE boot from UEFI to Legacy mode on the hardware server side. Thanks for your help.
[stack@director ~]$ openstack baremetal introspection bulk start
Setting available nodes to manageable...
Starting introspection of node: 70e3073a-a4e2-420c-aca9-586134e03e72
Starting introspection of node: b20644e3-2b18-4cd7-8dd0-cb1ad3818de6
Waiting for introspection to finish...
Introspection for UUID 70e3073a-a4e2-420c-aca9-586134e03e72 finished successfully.
Introspection for UUID b20644e3-2b18-4cd7-8dd0-cb1ad3818de6 finished successfully.