DHCP was enabled successfully when IP pool is exhausted

[Frank Wang]

In Pike, I created a subnet that disabled DHCP, then I create many VMs using this subnet, the IP pool in the subnet was exhausted. changing DHCP to true to update the subnet. there is no more IP in allocation_pools. the request succeeded. actually, DHCP agent can't create DHCP port to distribute IPs in the low level.

After I've looked up the bug record, I found that this kind of bug was reported and fixed about two years ago,https://review.openstack.org/#/c/174228/, but this fix was revert by this commit https://review.openstack.org/#/c/249060/

I'm wondering that this bug is still here， are we going to fix this kind of issue?

Thanks.
Frank.

[YAMAMOTO Takashi (yamamoto)]

looking at the code it seems still valid. i guess this can't be fixed without some kind of reservation mechanism.