Open Stack

Hi pierre,

I’m using a dedicated interface but this interface is the same for all
ironic networks inspection/provisioning/cleaning.

This interface works fine for inspection, my only issue is the pxe_filter
that the ironic inspector process allow during inspection correctly but
then tag as disallowed again at the end of the inspection, shouldn’t the
deploy process allow the Mac again before booting the node?

I can correctly see the conductor instruct the node to boot up using pxe
from the kvm console but the BootP process doesn’t load the IPA
kernel/initramfs as the dnsmasq pxe discard the request (because of the mac
being still tagged as «,ignore » within the hostdir file).

I’m a bit disappointed by this behavior.

Thanks a lot!

Le lun. 13 juin 2022 à 11:41, Pierre Riteau <pierre at stackhpc.com> a écrit :

> Hello Gaël,
>
> Which network_interface are you using for your nodes? Is your provisioning
> network different from the inspection network?
>
> Pierre
>
> On Mon, 13 Jun 2022 at 10:36, Gaël THEROND <gael.therond at bitswalk.com>
> wrote:
>
>> Hi everyone!
>>
>> I'm dealing with a strange issue today, we deployed IRONIC on a VICTORIA
>> platform, we activated the dnsmasq pxe filtering option at the inspector
>> level, it works great as only IRONIC listed hosts are then served by the
>> dnsmasq DHCP as those hosts are allowed using the dhcp hosts dir.
>>
>> BUT, I'm having a weird issue now.
>>
>> All my nodes are able to get an IP from the DHCP at the INSPECTION step,
>> however, as soon as the inspection step is finished, the mac related file
>> is once again filled with "<mac>,ignore" which prohibits further operations.
>>
>> This means as soon as I put that node as available and try to deploy an
>> "instance" on it (provision the host), it doesn't work as dnsmasq reject
>> the host boot DHCP requests.
>>
>> So, is there a way to instruct the ironic-conductor to edit/allow the
>> host the way the inspector is able to manipulate this file?
>>
>> Did I missed something?
>>
>> Thanks a lot!
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-discuss/attachments/20220613/ee23f324/attachment.htm>