Created attachment 913728 [details]
Description of problem:
Fail to start VM with error:
The host orange-vdsd.qa.lab.tlv.redhat.com did not satisfy internal filter Network because network(s) are missing.
Same even if the VM has no network
Create empty cluster policy solve the issue
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Create new VM
2. Start the VM
2014-07-01 06:36:25,942 WARN [org.ovirt.engine.core.bll.RunVmCommand] (ajp--127.0.0.1-8702-1) [1c97ffe2] CanDoAction of action RunVm failed. Reasons:VAR__ACTION__RUN,VAR__TYPE__VM,SCHEDULING_ALL_HOSTS_FILTERED_OUT,VAR__FILTERTYPE__INTERNAL,$hostName orange-vdsd.qa.lab.tlv.redhat.com,$filterName Network,$networkNames ,VAR__DETAIL__NETWORK_MISSING,SCHEDULING_HOST_FILTERED_REASON_WITH_DETAIL
It seems that VM fail to start only on AMD cluster with AMD host
Sounds related to cluster scheduling policy rather than network.
Also happened on Intel cluster when tried to migrate VM.
I can start Vms in my engine, is there anything specific about your display network configuration?
Can you reproduce this with DEBUG level enabled for org.ovirt.engine.core.bll.scheduling.policyunits.NetworkPolicyUnit ?
Display network is on ovirtmgmt (default).
engine with debug log attached.
Created attachment 917122 [details]
The issue was a configuration problem, for the network filter to allow a host, the host must have an interface with boot protocol.
The boot protocol for that host was set to "none".
Once changing the boot protocol there was no problem running a VM.
The referenced patch improves the reporting.
This is apparently caused due to Bug 1122475, leaving open for QE verification after that is fixed. Marking it as blocker since it doesn't allow running VMs while DHCP is configured.
Putting ON_QA as blocking vdsm bug is already VERIFIED. Martin, would you like to use this bug to also backport the logging change?
Verified on- oVirt Engine Version: 3.5.0-0.0.master.20140722232058.git8e1babc.el6
oVirt 3.5 has been released and should include the fix for this issue.