Description of problem: Packstack with Nova does not change the private interface to anything realistic on the host. Therefore, it fails in an ugly manner on CentOS 7 (host completed installation fine on CentOS 6.5, same packstack): ERROR : Failed to run remote script, stdout: Device eth1 does not exist stderr: Warning: Permanently added '10.102.34.175' (ECDSA) to the list of known hosts. + trap t ERR + ip addr show dev eth1 Device "eth1" does not exist. + echo Device eth1 does not exist + exit 1 ++ t ++ exit 1 [root@lgdrm920 Mgmt-OpenStack installation]# grep eth1 answers.txt CONFIG_NOVA_COMPUTE_PRIVIF=eth1 CONFIG_NOVA_NETWORK_PRIVIF=eth1 Version-Release number of selected component (if applicable): openstack-packstack-2014.1.1-0.25.dev1208.el7.noarch How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info: [root@lgdrm920 Mgmt-OpenStack installation]# ip addr show 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever 2: enp4s0f0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000 link/ether 00:1e:67:9f:35:92 brd ff:ff:ff:ff:ff:ff 3: enp4s0f1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000 link/ether 00:1e:67:9f:35:93 brd ff:ff:ff:ff:ff:ff 4: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000 link/ether 00:1e:67:69:77:74 brd ff:ff:ff:ff:ff:ff inet 10.102.34.175/20 brd 10.102.47.255 scope global dynamic eth0 valid_lft 157111sec preferred_lft 157111sec 5: eno1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN qlen 1000 link/ether 00:1e:67:69:77:75 brd ff:ff:ff:ff:ff:ff
I'm not sure what would be a good interface to choose from. In my case, either enp4s0f0 or enp4s0f1 would have been good.
For whatever it's worth, I'm getting the same error, and ... : [rbowen@localhost ~]$ sudo ip addr show 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever inet6 ::1/128 scope host valid_lft forever preferred_lft forever 2: enp0s25: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast state DOWN qlen 1000 link/ether f0:de:f1:dc:2a:08 brd ff:ff:ff:ff:ff:ff 3: wlp3s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000 link/ether 10:0b:a9:c7:05:60 brd ff:ff:ff:ff:ff:ff inet 192.168.0.176/24 brd 192.168.0.255 scope global dynamic wlp3s0 valid_lft 85484sec preferred_lft 85484sec inet6 fe80::120b:a9ff:fec7:560/64 scope link valid_lft forever preferred_lft forever So it looks like, in this case, I'd want wlp3s0. This is a fresh CentOS7 install. Perhaps packstack needs to be a little smarter than just having default interface names? Looks possibly related to the discussion at https://openstack.redhat.com/forum/discussion/500/rdo-install-issues-on-f19-eth1-django14/p1
If "packstack --allinone" is failing, that is an actual problem and we should try to correct it. But if you are encountering this error on a multi-node install, I'm not sure that there's any sane way for packstack to select among multiple interfaces. The solution in this latter case is to run "packstack --gen-answer-file answers.txt" and then edit the answers file appropriately before running packstack. Can you confirm whether or not this was a --allinone install?
I was not using an allinone. But I think a reasonable hidden assumption is that it should be the correct value of the node I'm running packstack on.
If you're not performing an all-in-one install you are already explicitly setting a number of configuration options. Furthermore, in a system with multiple interfaces -- as in your example, or in rbowen's example -- there is *no way* for Packstack to intelligently select which of the available interfaces should be used for this option. I think it makes sense that we require CONFIG_NOVA_NETWORK_PRIVIF to be set explicitly.