Bug 1390557 - Migrating "Powering Up" VM will drop the VM to the "pause" state
Summary: Migrating "Powering Up" VM will drop the VM to the "pause" state
Keywords:
Status: CLOSED INSUFFICIENT_DATA
Alias: None
Product: vdsm
Classification: oVirt
Component: Core
Version: 4.18.15
Hardware: x86_64
OS: Linux
unspecified
high vote
Target Milestone: ---
: ---
Assignee: Dan Kenigsberg
QA Contact: Artyom
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2016-11-01 11:56 UTC by Artyom
Modified: 2016-11-10 11:44 UTC (History)
5 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2016-11-10 11:44:18 UTC
oVirt Team: Virt
rule-engine: planning_ack?
rule-engine: devel_ack?
rule-engine: testing_ack?


Attachments (Terms of Use)
logs (159.12 KB, application/zip)
2016-11-01 11:56 UTC, Artyom
no flags Details

Description Artyom 2016-11-01 11:56:41 UTC
Created attachment 1216071 [details]
logs

Description of problem:
I can see this bug only in automation run.
Test start the VM and when the VM has state "Powering Up" migrates it.

Version-Release number of selected component (if applicable):
vdsm-yajsonrpc-4.18.15.2-1.el7ev.noarch
vdsm-jsonrpc-4.18.15.2-1.el7ev.noarch
vdsm-hook-vmfex-dev-4.18.15.2-1.el7ev.noarch
vdsm-hook-openstacknet-4.18.15.2-1.el7ev.noarch
vdsm-api-4.18.15.2-1.el7ev.noarch
vdsm-infra-4.18.15.2-1.el7ev.noarch
vdsm-hook-vhostmd-4.18.15.2-1.el7ev.noarch
vdsm-python-4.18.15.2-1.el7ev.noarch
vdsm-cli-4.18.15.2-1.el7ev.noarch
vdsm-4.18.15.2-1.el7ev.x86_64
vdsm-hook-ethtool-options-4.18.15.2-1.el7ev.noarch
vdsm-xmlrpc-4.18.15.2-1.el7ev.noarch
vdsm-hook-fcoe-4.18.15.2-1.el7ev.noarch

ovirt-engine-4.0.5.4-0.1.el7ev.noarch

How reproducible:
10% only under automation tests

Steps to Reproduce:
1. Start the VM
2. Migrate the VM in "Powering Up" state
3.

Actual results:
Migration succeeds, but the VM appears in the pause state on the destination host

Expected results:
Migration succeeds, and the VM has staet "Up"

Additional info:
I do not sure if the bug relates to the engine or vdsm

Comment 1 Dan Kenigsberg 2016-11-01 12:34:33 UTC
Is this a regression? did the same test pass with vdsm-v4.18.2 ?

What happens if you manually continue the VM on destination? Does the guest start running all right?

Comment 2 Artyom 2016-11-01 13:20:55 UTC
Like I said, it happens to me only under automation run(not in the same test case), so I can not reproduce it manually.
I believe it regression because the bug appears only in last automation runs, but again I do not sure for 100% because from all tests that we run it happens only for some single test case.

Comment 3 Dan Kenigsberg 2016-11-01 13:29:18 UTC
Artyom, what is the most recent version of vdsm that passes this test %100? Can you stop the test after failure, in order to see what happens inside the guest and whether manual "continue" solves things?

Comment 4 Artyom 2016-11-01 14:01:46 UTC
We do not have such long history of runs under Jenkins, but from email, I can see that 4.0.4-1 does not have tests that fail because of this bug.
I will try to catch the bug on my local automation environment and will check the guest, but I do not sure if I will succeed.
Until then can someone take a look at logs?

Comment 5 Tomas Jelinek 2016-11-08 12:16:42 UTC
hmmm, I have not found anything too interesting in the logs, could you please try to provide the qemu and libvirt logs from the destination host?
thanx

Comment 6 Artyom 2016-11-10 10:55:31 UTC
I do not have libvirt and qemu logs for the automation run.
I tried to reproduce this problem locally, but without result
100 iterations:
1) Start the VM
2) Migrate the VM straight forward after start
3) Stop the VM

Maybe this problem relates to some problematic host

I believe you can close it with insufficient data, in case if I will see this error again I will attach logs.

Comment 7 Tomas Jelinek 2016-11-10 11:44:18 UTC
Thank you!
So closing as insufficient data since without libvirt/qemu logs there is not much we can do.

If it will happen again and you will have this logs please reopen.


Note You need to log in before you can comment on or make changes to this bug.