Description of problem: when deploying pre-deployed server with undercloud_public_host set to IPv6 address, os-collect config does not parse IPv6 address from the url correctly due to the regex expression used: sed -e 's|http.*://\([^:]*\).*|\1|' which gets only the first octet of the IPv6 address (example: METADATA_IP='[2001'). I would assume that even after that is fixed the ping will still fail because only ping (not ping6 or -6 parameter is used) is then triggered: echo -n 'Trying to ping metadata IP [2001...' os-collect-config[12730]: + local COUNT=0 os-collect-config[12730]: + ping -c 1 '[2001' os-collect-config[12730]: + COUNT=1 This also eventually results in fall back to dhcp-all-interfaces with --cleanup parameter called that removes statically configured interfaces configuration. Version-Release number of selected component (if applicable): OSP13 puddle: 2019-03-18.1 How reproducible: 100% Steps to Reproduce: 1. Set undercloud_public_host = <ipv6 address> 2. run openstack undercloud install 3. trigger overcloud deploy for pre-deployed nodes Actual results: Deployment fails with network configuration, interfaces get generic configuration (ifcfg-ethX): # This file is autogenerated by os-net-config DEVICE=eth3 ONBOOT=yes HOTPLUG=no NM_CONTROLLED=no BOOTPROTO=dhcp Expected results: Deployment continues correctly with static configuration preserved Additional info:
switching component. the bug here is in network/scripts/run-os-net-config.sh
Works as expected now. Tested on puddle 2019-04-10.1
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2019:0939