clustervision
iPXE DHCP failure requiring manual ifopen on specific HPE node
Hi,
We have over a hundred compute nodes that are provisioned by TrinityX 15.1, consisting mostly of HPE servers with some Dell servers. The system has been working well overall, but we're experiencing a PXE boot issue with one specific HPE node that requires manual intervention in the iPXE shell.
Issue Summary: The node cannot complete automatic PXE boot with Luna, but can boot successfully when using manual iPXE commands.
Detailed Problem Description:
Initial PXE Stage: The node successfully obtains an IP address and downloads luna_ipxe.efi, but then fails because it expects to find autoexec.ipxe, which is not present. This was confirmed in the host message logs.
iPXE Stage: After it fail to locate autoexec.ipxe, iPXE (version 1.21.1) starts and attempts to open interface net1 and obtain an IP from the controller. However, this DHCP request fails, causing the system to attempt PXE boot from alternative interfaces (such as InfiniBand).
Manual Workaround: When I interrupt the iPXE stage and manually execute the following commands, the boot process succeeds:
ifopen net1 dhcp imgfetch http://
:7051/boot imgstat boot boot
This sequence successfully boots into the Luna boot menu.
Is there a recommended workaround or configuration change for nodes that require manual interface initialization?
Thanks a lot!
We are still stuck with this problem, any suggestion would be appreciated. Thanks!
Hi chiensh. At first sight this seems to be a very specific problem. Can you share a bit more details about the platform (hpe?) and the nic? Probably not the answer you're looking for, but did you try upgrading the nics firmware?
Yes, the problem should be due to the nic firmware, and we will change the network card, it seems to be the faster solution to this.
https://github.com/ipxe/ipxe/issues/1548#issuecomment-3379578484