Masternode works incorrectly when additional interface with DHCP configured
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Fix Committed
|
Medium
|
Sergii Golovatiuk | ||
5.0.x |
Fix Committed
|
Medium
|
Sergii Golovatiuk |
Bug Description
If masternode will have an additional interface (beside eth0) with DHCP configured then after reboot masternode using DNS servers from this DHCP interface and will not configure local DNS server to use. So after the restart ability to go on nodes by names, node-* mask will disappear.
How to reproduce:
1. Install masternode with at least two interfaces and for simplifying second interface should be connected to some network with DHCP server (usually it's kind of administrative network). After masternode will complete the installation reboot it and try to boot up some nodes.
2. Provision and deploy any openstack variant and try to ssh to any of the nodes using name (node-*, controller-* and so on). You wouldn't be able to do that.
Changed in fuel: | |
importance: | Undecided → Medium |
assignee: | nobody → Fuel Library Team (fuel-library) |
milestone: | none → 5.1 |
What you describe is a normal behavior when you have several DHCP clients. However, Fuel has several techniques such as dhcp-enter-hooks. By normal circumstances it shouldn't allow to overwrite /etc/resolv.conf even if you have dhcp enabled for some interfaces.
https:/ /github. com/stackforge/ fuel-library/ blob/master/ deployment/ puppet/ nailgun/ manifests/ host.pp# L54