Bug 1093394

Summary: RFE: Be more forgiving with disk/net poll failures
Product: [Community] Virtualization Tools Reporter: Dean Hunter <deanhunter>
Component: virt-managerAssignee: Pavel Hrdina <phrdina>
Status: CLOSED NEXTRELEASE QA Contact:
Severity: medium Docs Contact:
Priority: unspecified    
Version: unspecifiedCC: berrange, crobinso, gscrivan, phrdina, rbalakri, virt-maint
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2017-09-11 07:39:11 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:
Embargoed:
Attachments:
Description Flags
virt-manager --debug &>virt-manager.log &
none
./virt-manager --debug &>../virt-manager.log & none

Description Dean Hunter 2014-05-01 15:20:09 UTC 
Description of problem:

I create, destroy and recreate virtual machines guests frequently using virt-install and virsh commands in scripts.  I watch the virt-manager graphs to gauge the activity of the guests as many of the commands are asynchronous.  I have noticed for some time that often the graphs do not display for a recreated guest.  It is necessary to stop and start virt-manager to display the graphs.  The graphs are almost always displayed for a new guest.


Version-Release number of selected component (if applicable):

Installed Packages
virt-manager.noarch                 1.0.1-1.fc20                  @local-updates


How reproducible: more than 50% of the time


Steps to Reproduce:

1. virt-install
2. virsh start
3. virsh stop
4. virsh undefine
5. virt-install


Actual results: 

The new guest is displayed in virt-manager but the graphs are empty.


Expected results:

The new guest is displayed in virt-manager and the graphs start scrolling.


Additional info:
Comment 1 Cole Robinson 2014-05-01 16:20:03 UTC 
Thanks for the report. Please provide virt-manager --debug output when reproducing
Comment 2 Dean Hunter 2014-05-01 16:47:37 UTC 
Created attachment 891586 [details]
virt-manager --debug &>virt-manager.log &
Comment 3 Cole Robinson 2014-05-02 14:29:32 UTC 
Hmm, it's still a bit unclear what's causing the issue. Can you reproduce with virt-manager upstream which has some improved debugging?

git clone git://git.fedorahosted.org/virt-manager.git
cd virt-manager
./virt-manager --debug
Comment 4 Dean Hunter 2014-05-03 17:45:54 UTC 
Created attachment 892161 [details]
./virt-manager --debug &>../virt-manager.log &
Comment 5 Cole Robinson 2014-05-03 18:01:59 UTC 
Hmm, this trace is slightly different but looks more expected.

The gist of the issue is that when a VM is shutting down or starting up or migrating, the disk polling bits often fail. Once a disk poll fails, virt-manager blacklists that disk device and no longer polls it, even if the VM reboots.

We should make this heuristic a little more forgiving, maybe require the poll to fail 3 times in a row before we give up on it, and consider retrying the poll if a VM reboots.
Comment 6 Dean Hunter 2014-05-03 19:16:53 UTC 
Polling fails, hmmm... 

Is this related to why sometimes all the graphs stop moving when I request the shutdown of one guest?
Comment 7 Cole Robinson 2014-05-03 20:46:28 UTC 
(In reply to Dean Hunter from comment #6)
> Polling fails, hmmm... 
> 
> Is this related to why sometimes all the graphs stop moving when I request
> the shutdown of one guest?

Very likely
Comment 8 Pavel Hrdina 2017-09-08 08:33:48 UTC 
Upstream patch posted:

https://www.redhat.com/archives/virt-tools-list/2017-September/msg00058.html
Comment 9 Pavel Hrdina 2017-09-11 07:39:11 UTC 
Upstream commit:

commit 12117ba148ec47eb2aa15e192c6026a2c3026ed1
Author: Pavel Hrdina <phrdina>
Date:   Fri Sep 8 09:36:58 2017 +0200

    connection: change blacklist from array to dict