<div dir="ltr">Hello, I'm having an issue right now, or maybe it's a configuration error.<br><br>Here's my problem, ( I'm using a fully deployed opentstack grizzly env with segmented networks<br>and one cc controlling XX number of computes clustered )<br>
<br>All of that works fine, we love openstack. One thing though we are trying to work around<br>is the dns / dhcp issue. When you run an env as such, not using some form of internal dns is pretty insane.<br>Whether it's a puppet server or a monitoring tool, dns is essential. I've tried a few setups<br>
trying to use an external DHCP server, but seems this is not really an option. ( Perhaps Havana release changed some things, just haven't had time to check it out. But have yet to find a successful document )<br><br>What I've decided to do as a simple solution for now ( or I thought it would be ) is<br>
simply use an internal dns server running (bind probably, or mydns), and sending the compute nodes dnsmasq process to the other dns server for requests. So openstack will handle the dhcp, but send the dns to the server I want which is fine with me, we have internal scripts for dynamic updates unrelated. I'm wondering if this setting is hard coded some where,<br>
I've been digging but haven't found anything.<br><br>I've specified this simple setting in the nova.conf of the compute node ( which I thought should work ),<br><br>#dns tests<br>dhcp_domain=devtest<br>dhcp_lease_time=480<br>
dns_server=10.5.20.15<br><br>Why I thought perhaps it's hard coded some where to use the host ip is because of I grep the process after restarting it, this is the output.<br><br>"/usr/sbin/dnsmasq --strict-order --bind-interfaces --conf-file= --pid-file=/var/lib/nova/networks/nova-br101.pid --listen-address=10.5.20.3 --except-interface=lo --dhcp-range=set:int-net,10.5.20.2,static,255.255.255.0,480s --dhcp-lease-max=256 --dhcp-hostsfile=/var/lib/nova/networks/nova-br101.conf --dhcp-script=/usr/bin/nova-dhcpbridge --leasefile-ro --domain=alchint --no-hosts --addn-hosts=/var/lib/nova/networks/nova-br101.hosts --no-resolv --server=10.5.20.15<br>
<br>It shows it should be sending requests to the server, but the vm will still get the compute nodes IP for nslookups. Although the vm will display the correct domain name, so it passes some information.<br><br>I've looked through "/usr/share/pyshared/nova/network" manager.py but couldn't find an explicit<br>
declaration. Not sure if I'm taking the wrong approach, and there's a simple error. As the documentation is pretty lacking, we've had to figure out most of it. I'm in the process of creating an individual dns_masq.conf custom file to specify to nova..(maybe the flags it provides on default conflict with my desired result) Just wondering if someone had this issue.<br>
<br>Thanks in advance for any assistance / guidance in the right direction, we're proud active members of the open source community.<br><br><br>Regards<br><br><br></div>