Bug 1153128
Summary: | Cannot start nova-network on juno - Centos7 | ||
---|---|---|---|
Product: | [Community] RDO | Reporter: | Bruno Bompastor <b.bompastor> |
Component: | openstack-packstack | Assignee: | Martin Magr <mmagr> |
Status: | CLOSED EOL | QA Contact: | Ofer Blaut <oblaut> |
Severity: | high | Docs Contact: | |
Priority: | unspecified | ||
Version: | unspecified | CC: | aortega, apevec, beagles, berrange, dasmith, derekh, eglynn, gchamoul, rbryant, sbauza, sferdjao, sgordon, srevivo, vromanso |
Target Milestone: | --- | ||
Target Release: | Juno | ||
Hardware: | x86_64 | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2016-05-19 15:54:40 UTC | Type: | Bug |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Bruno Bompastor
2014-10-15 16:11:20 UTC
Basically the parameter flat_interface=lo is wrong. If you comment this parameter on nova.conf it takes 'lo' as default. The only way to overcome this is by forcing empty string for the interface on the code. interface = '' if interface: msg = _('Adding interface %(interface)s to bridge %(bridge)s') LOG.debug(msg, {'interface': interface, 'bridge': bridge}) out, err = _execute('brctl', 'addif', bridge, interface, check_exit_code=False, run_as_root=True) if (err and err != "device %s is already a member of a bridge; " "can't enslave it to bridge %s.\n" % (interface, bridge)): msg = _('Failed to add interface: %s') % err raise exception.NovaException(msg) out, err = _execute('ip', 'link', 'set', interface, 'up', check_exit_code=False, run_as_root=True) I can also reproduce this on: openstack-nova-network.noarch 2014.2-0.7.rc2.el7.centos @openstack-juno I opened a bug (https://bugs.launchpad.net/nova/+bug/1376596) before on upstream but it was rejected as a misconfiguration. So maybe it should be changed on the puppet module... What's the conclusion here? Is it nova, puppet-nova or packstack issue? AFAICT this only affects allinone, so workaround could be force neutron networking for allinone? I think the fix should come from nova.... But if that is not the case packstack should put in place the workaround on this bug https://bugzilla.redhat.com/show_bug.cgi?format=multiple&id=1153079 to fix this. Basically create a dummy interface and fix the allinone configuration to change the loopback to the newly created dummy interface. The bug is in how packstack configures allinone for nova networking. The private interfaces should be empty for allinone. Bridging lo is kind of bogus and completely unnecessary for allinone. I believe this should be reassigned to packstack. Ah, just realized it IS assigned to packstack. An additional note, the 'lo' is set in single_step_aio_install() in installer/run_setup.py. Unfortunately, just changing or removing those lines doesn't do what's required - it reverts to the internal default of eth1. As an aside, you cannot set it the *_PRIVIF to an empty value in the answer file either - which is wrong. A more successful approach would be for the PRIVIF's to default to empty and then only validate for empty if multiple compute nodes are specified - in which event not providing an actual physical network for "private" tenant traffic would be an error. Added new upstream BZ against packstack. This bug was accidentally moved from POST to MODIFIED via an error in automation, please see mmccune with any questions This bug is against a Version which has reached End of Life. If it's still present in supported release (http://releases.openstack.org), please update Version and reopen. |