Description of problem: Installing puddle 2015-07-17 in a virtualized environment and running introspection, I get an error (after a long timeout) that all node discoveries failed: [stack@instack ~]$ openstack baremetal introspection bulk start Setting available nodes to manageable... Starting introspection of node: dddcccc8-4586-471c-a9f0-cad0ccddf339 Starting introspection of node: 3535c1b0-687d-4895-860d-a641e7536aef Starting introspection of node: 9f4c43da-be9a-47cd-8d1b-b5ab97473b9c Starting introspection of node: 293864c6-1dfb-455f-b2d3-b8fd10c119dc Starting introspection of node: 9c33b848-e972-418c-b19e-b073df8f828a Starting introspection of node: ca292f2d-1299-44e3-b8f8-7619ffc5aed5 Starting introspection of node: fd8b2460-4392-424a-9f2b-22f8c996ae99 Starting introspection of node: fd3d5426-f069-4de3-bcf6-d69445d717b4 Waiting for discovery to finish... ERROR: rdomanager_oscplugin.utils.wait_for_node_discovery Discovery didn't finish for nodes dddcccc8-4586-471c-a9f0-cad0ccddf339,3535c1b0-687d-4895-860d-a641e7536aef,9f4c43da-be9a-47cd-8d1b-b5ab97473b9c,293864c6-1dfb-455f-b2d3-b8fd10c119dc,9c33b848-e972-418c-b19e-b073df8f828a,ca292f2d-1299-44e3-b8f8-7619ffc5aed5,fd8b2460-4392-424a-9f2b-22f8c996ae99,fd3d5426-f069-4de3-bcf6-d69445d717b4 Setting manageable nodes to available... Node dddcccc8-4586-471c-a9f0-cad0ccddf339 has been set to available. Node 3535c1b0-687d-4895-860d-a641e7536aef has been set to available. Node 9f4c43da-be9a-47cd-8d1b-b5ab97473b9c has been set to available. Node 293864c6-1dfb-455f-b2d3-b8fd10c119dc has been set to available. Node 9c33b848-e972-418c-b19e-b073df8f828a has been set to available. Node ca292f2d-1299-44e3-b8f8-7619ffc5aed5 has been set to available. Node fd8b2460-4392-424a-9f2b-22f8c996ae99 has been set to available. Node fd3d5426-f069-4de3-bcf6-d69445d717b4 has been set to available. Discovery completed. Version-Release number of selected component (if applicable): python-rdomanager-oscplugin-0.0.8-41.el7ost.noarch How reproducible: 100% Steps to Reproduce: 1. Install according to the guide until you get to the introspection phase Actual results: I tried it twice in a row and it seems to fail consistently. After the failure the nodes are left in "power on" state: [stack@instack ~]$ ironic node-list +-------+------+---------------+-------------+-----------------+-------------+ | UUID | Name | Instance UUID | Power State | Provision State | Maintenance | +-------+------+---------------+-------------+-----------------+-------------+ | dddcc | None | None | power on | available | False | | 3535c | None | None | power on | available | False | | 9f4c4 | None | None | power on | available | False | | 29386 | None | None | power on | available | False | | 9c33b | None | None | power on | available | False | | ca292 | None | None | power on | available | False | | fd8b2 | None | None | power on | available | False | | fd3d5 | None | None | power on | available | False | +-------+------+---------------+-------------+-----------------+-------------+ I made sure to turn all nodes off before retrying.
It looks like Ironic is able to manage the power for the virtual machines, but the ramdisk hung at some point. We need to look on the console of the virtual machines to check what happened. Does this setup have multiple NICs on the provisioning network? If so, this could be a duplicate of https://bugzilla.redhat.com/show_bug.cgi?id=1243109
How do I check if the setup has multiple NICs on the provisioning network?
Udi, Were you deploying in a virt environment with multiple bridges? If so, I think this is a duplicate of https://bugzilla.redhat.com/show_bug.cgi?id=1243109
(In reply to Udi from comment #4) > How do I check if the setup has multiple NICs on the provisioning network? You would have had to use this before running instack-virt-setup: export TESTENV_ARGS="--baremetal-bridge-names 'brbm brbm1 brbm2'" I know that's documented, but it has known issues, hence the BZ above.
Yes I ran export TESTENV_ARGS="--baremetal-bridge-names 'brbm brbm1 brbm2'". Should we skip this step for now ?
(In reply to Udi from comment #7) > Yes I ran export TESTENV_ARGS="--baremetal-bridge-names 'brbm brbm1 brbm2'". > Should we skip this step for now ? Yes, you should skip this for now. There is a known issue which hasn't been properly triaged yet. The purpose of multiple bridges is to test bonding in virt. If you just use the default settings for instack-virt-setup, you can use the single-nic-vlans templates to set up network isolation over a single NIC.
*** This bug has been marked as a duplicate of bug 1243109 ***