Bug 1372029

Summary: overcloud deployment fails with no valid hosts found
Product: Red Hat OpenStack Reporter: Dave Wilson <dwilson>
Component: rhosp-directorAssignee: Dmitry Tantsur <dtantsur>
Status: CLOSED INSUFFICIENT_DATA QA Contact: Omri Hochman <ohochman>
Severity: medium Docs Contact:
Priority: unspecified    
Version: 9.0 (Mitaka)CC: chih-hsien.chien, dbecker, dtantsur, dwilson, jcoufal, krzysztofx.malkowski, mburns, morazi, rhel-osp-director-maint, tvignaud
Target Milestone: ---Keywords: ZStream
Target Release: 9.0 (Mitaka)Flags: dtantsur: needinfo? (dwilson)
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2017-01-19 15:02:22 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:
Attachments:
Description Flags
associated log files none

Description Dave Wilson 2016-08-31 18:08:42 UTC
Created attachment 1196485 [details]
associated log files

Description of problem: When deploying an overcloud of 18 computes with only 18 available compute ironic nodes, deployment fails. "nova list" show 3 nodes in error state. "ironic node-list" show non-existent "Instance UUID" attached to a node with a provisioning state "available". Note, a deployment of 18 computes with 36 available ironic compute nodes deploys and using node binding via capabilities also incurs no issues. For the instances  in "error" state the logs reveal "NoValidHost: No valid host was found. There are not enough hosts available." Before deallocating instance. Attached are the output from nova list and ironic node-list, compute.log and conductor.log


Version-Release number of selected component (if applicable):
python-openstackclient-2.2.0-1.el7ost.noarch
openstack-tripleo-0.0.8-0.2.d81bd6dgit.el7ost.noarch


How reproducible: 100%


Steps to Reproduce:
1.Deploy overcloud of 18 computes with only 18 availble ironic "compute" nodes
2.Observe failure of a few compute instances

Actual results:Deplyoment fails as described above


Expected results: A successful overcloud deployment


Additional info:

Comment 2 Dmitry Tantsur 2016-10-17 08:42:13 UTC
Which OSP version is it? Judging by openstackclient version it is OSP9, right?

> "ironic node-list" show non-existent "Instance UUID" attached to a node with a provisioning state "available"

What do you mean by non-existent? You don't see these instances in nova show? Is it possible that they are left from a previous deployment that failed to clean up properly?

Comment 3 Dmitry Tantsur 2017-01-19 15:02:22 UTC
Please reopen when you're ready to provide the missing information. Also please make sure to check http://docs.openstack.org/developer/ironic/deploy/troubleshooting.html#nova-returns-no-valid-host-was-found-error