maas-devel team mailing list archive

[Raphaël Badin <raphael.badin@xxxxxxxxxxxxx>]

On 02/07/2014 12:53 AM, Julian Edwards wrote:
> The dhcp probe task is still causing the dhcp config to get rewritten on my
> box.  This is using the very latest daily build:
> 1.4+bzr1853+dfsg+1898+231~ppa0~ubuntu14.04.1
>
> /var/log/maas/celery.log:
> [2014-02-07 09:48:17,151: INFO/MainProcess] Task
> provisioningserver.tasks.periodic_probe_dhcp[e96c29f4-ba0d-48ee-
> b09d-8587f94d7dc7] succeeded in 3.513536813s: None
> [2014-02-07 09:48:17,961: INFO/MainProcess] Task
> provisioningserver.tasks.write_dhcp_config[171180a4-
[...]

That's odd: I tested this on canonistack yesterday and didn't see that. 
 I'll test again today on my microservers.

> I'm also seeing the region celery log doing something similar:
>
> /var/log/maas/celery-region.log:
> [2014-02-07 09:50:45,122: INFO/MainProcess] Task
[...]

Yes, we are aware of this problem: we also need to fix the DNS-related 
signals so that changes to the foreign_dhcp field on a interface don't 
trigger a DNS config rewrite.  We were planning to fix this the same way 
the DHCP config rewrite was fixed but let's QA this first fix before we 
apply the same trick to the DNS config rewrite.

> There is only my MAAS DHCP server on the network, which is ignored.  Maybe
> that triggers a bug in the field change code?

So far I've done my testing on canonistack; the test I'm planning to do 
today, with my microservers, will use a setup similar to yours.  I'll 
report back when I'm done.

R.