Comment 35 for bug 1707999
Revision history for this message
| Blake Rouse (blake-rouse) wrote Re: pod VM fails to PXE boot after receiving multiple DHCP offers from both primary and secondary rack controllers, for different IPs | #35 |
Jason,
I think your theory is correct.
Context on how it works:
We do write the hostmaps into the config but dhcpd is not reloaded (isc-dhcpd doesn't support reload) we using the OMAPI to update the hostmaps so no to bounce the DHCP server. The hostmap is only written to the dhcpd.conf just incase the leases file is deleted and dhcpd is restarted.
The updating of this information is out of band, from the machine being turned on. Meaning that MAAS does not block until hostmap is written before powering the machine on.
We will will need to look into how we can verify that its written before actually powering the machine on. I also think you are correct in that normal machines do not have this problem because they take long enough to power up before it makes its first PXE request.
Really thanks for digging into this and figuring out what was really going wrong. The hard part now is fixing it.
Thanks!