Hi Bryan, If the steps I sent you work, then please let me know and I can write up a patch to add pxe_wol to Apex.
Tim Rozet Red Hat SDN Team ----- Original Message ----- From: "BRYAN L SULLIVAN" <[email protected]> To: "Tim Rozet" <[email protected]> Cc: [email protected] Sent: Monday, January 23, 2017 1:57:53 PM Subject: RE: [opnfv-tech-discuss] [Apex] Danube install help Yes, the src reference was an error in my notes. I'll try again with the notes you added. Not supporting PXE_WOL significantly reduces the types of machines that we can use for OPNFV development. I recommend that all bare-metal installers consider that PXE_WOL may be required, and provide a deploy parameter and hook at the proper time so that the PXE_WOL wakeup can be sent by the script to the power address (MAC) referenced in the lab config. At the very least if PXE_WOL is indicated as the power-type in the lab config, the install script should pause, inform the user that they need to wakeup a specific (or all) nodes at that time, and proceed when the user hits enter. Thanks, Bryan Sullivan | AT&T -----Original Message----- From: Tim Rozet [mailto:[email protected]] Sent: Monday, January 23, 2017 7:35 AM To: SULLIVAN, BRYAN L <[email protected]> Cc: [email protected] Subject: Re: [opnfv-tech-discuss] [Apex] Danube install help Hi Bryan, The package requirements should be the same. One problem I see from your output is you are trying to install source rpms. The one thing that has changed are the network settings format. We added a lot of comments there so hopefully that all makes sense. To use a single network, just only enable the first network in the network settings. There is no need to use the deprecated '--flat' argument anymore. The deployment will also use a single compute node by default. You can take the latest 4.0 artifacts, they should be stable. For pxe_wol, you will need to interrupt the deployment and make a couple changes as we dont currently support pxe_wol: 1. add at https://github.com/opnfv/apex/blob/master/lib/overcloud-deploy-functions.sh#L289 : cat > deploy_command << EOF openstack overcloud deploy --templates $DEPLOY_OPTIONS --timeout 90 EOF exit 0 2. login to undercloud: opnfv-util undercloud 3. sudo -i; vi /etc/ironic/ironic.conf 4. set enabled drivers: enabled_drivers = pxe_ipmitool,pxe_ssh,pxe_drac,pxe_ilo,pxe_wol 5. systemctl restart openstack-ironic-conductor;exit 6. cd /home/stack; . stackrc; vi instackenv.json (this is your parsed inventory file for OOO) 7. for each "pm_type", change it to be "pxe_wol" 8. for each "pm_addr", change it to be the broadcast address of your ctlplane subnet (example 192.0.20/24 would be 192.0.2.255) 9. add "pm_port", under each "pm_addr" and set it to be the dest port for your WOL setup, default is 9 10. Set global bash var: dns_server_ext="--nameserver <your DNS server>" 11. Do the following commands: for flavor in baremetal control compute; do if openstack flavor list | grep ${flavor}; then openstack flavor delete ${flavor} fi openstack flavor create --id auto --ram 4096 --disk 39 --vcpus 1 ${flavor} done openstack baremetal import --json instackenv.json openstack baremetal configure boot openstack flavor set --property "cpu_arch"="x86_64" --property "capabilities:boot_option"="local" baremetal openstack flavor set --property "cpu_arch"="x86_64" --property "capabilities:boot_option"="local" --property "capabilities:profile"="control" control openstack flavor set --property "cpu_arch"="x86_64" --property "capabilities:boot_option"="local" --property "capabilities:profile"="compute" compute neutron subnet-update $(neutron subnet-list | grep -Ev "id|tenant|external|storage" | grep -v \\\\-\\\\- | awk {'print $2'}) ${dns_server_ext} 12. cat deploy_command 13. Copy the output of step 12 and execute in terminal Ping me on #opnfv-apex if something doesnt work. Tim Rozet Red Hat SDN Team ----- Original Message ----- From: "BRYAN L SULLIVAN" <[email protected]> To: [email protected] Sent: Saturday, January 21, 2017 7:41:17 PM Subject: Re: [opnfv-tech-discuss] [Apex] Danube install help Also, it’s clear that something has changed re the dependency upon jinja2, python-3-ipmi, etc … per the error I got below when trying to install the RPMs: --> Processing Dependency: python3-ipmi for package: --> opnfv-apex-common-4.0-20170121.noarch --> Processing Dependency: python3-jinja2 for package: --> opnfv-apex-common-4.0-20170121.noarch --> Finished Dependency Resolution Error: Package: opnfv-apex-common-4.0-20170121.noarch (/opnfv-apex-common-4.0-20170121.noarch) Requires: python3-jinja2 Error: Package: opnfv-apex-common-4.0-20170121.noarch (/opnfv-apex-common-4.0-20170121.noarch) Requires: python3-ipmi You could try using --skip-broken to work around the problem You could try running: rpm -Va --nofiles –nodigest [bryan@opnfv apex]$ sudo yum install -y python-jinja2-2.8-5.el7.centos.src.rpm python-markupsafe-0.23-9.el7.centos.src.rpm python3-ipmi-0.3.0-1.noarch.rpm Loaded plugins: fastestmirror, langpacks Loading mirror speeds from cached hostfile * base: mirror.hostduplex.com * epel: mirrors.kernel.org * extras: mirror.scalabledns.com * updates: mirrors.kernel.org No package python-jinja2-2.8-5.el7.centos.src.rpm available. No package python-markupsafe-0.23-9.el7.centos.src.rpm available. No package python3-ipmi-0.3.0-1.noarch.rpm available. Error: Nothing to do [bryan@opnfv apex]$ Thanks, Bryan Sullivan | AT&T From: SULLIVAN, BRYAN L Sent: Saturday, January 21, 2017 4:36 PM To: [email protected] Subject: RE: [Apex] Danube install help One other thing I forgot in the list added below. Thanks, Bryan Sullivan | AT&T From: SULLIVAN, BRYAN L Sent: Saturday, January 21, 2017 4:33 PM To: [email protected] Subject: [Apex] Danube install help Hi Apex team, I want to get going on installing Apex in a virtual or non-HA/3-node (jumphost, control, compute) environment, for Copper, Models, VES. I need to know how the install process has changed for Danube, e.g. pre-reqs and command line options. Here is the process I used for Colorado virtual install. Is all this still correct? · sudo yum install -y epel-release · sudo yum install -y python34 · sudo yum install -y https://www.rdoproject.org/repos/rdo-release.rpm · sudo yum install -y python-jinja2-2.8-5.el7.centos.src.rpm python-markupsafe-0.23-9.el7.centos.src.rpm python3-ipmi-0.3.0-1.noarch.rpm · Download latest Apex RPMs, e.g. ( How do we know which build is stable enough to test with? Digging through Jenkins is not a great solution…) o wget http://artifacts.opnfv.org/apex/opnfv-apex-4.0-20170121.noarch.rpm o wget http://artifacts.opnfv.org/apex/opnfv-apex-common-4.0-20170121.noarch.rpm o wget http://artifacts.opnfv.org/apex/opnfv-apex-undercloud-4.0-20170121.noarch.rpm · sudo yum install -y opnfv-*.rpm · sudo opnfv-deploy -v -d /etc/opnfv-apex/os-nosdn-nofeature-noha.yaml -n /etc/opnfv-apex/network_settings.yaml · sudo opnfv-util undercloud And for the non-HA bare metal install with flat networking and PXE_WOL, what are the options I need to include for the deploy? If you’ve documented all this for Danube, a link to the docs/wiki would help. Thanks, Bryan Sullivan | AT&T _______________________________________________ opnfv-tech-discuss mailing list [email protected] https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss _______________________________________________ opnfv-tech-discuss mailing list [email protected] https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss
