Bug 728603 - guest may remain in existence on dst host if src libvirtd stopped during tunneled migration
Summary: guest may remain in existence on dst host if src libvirtd stopped during tunn...
Keywords:
Status: ASSIGNED
Alias: None
Product: Virtualization Tools
Classification: Community
Component: libvirt
Version: unspecified
Hardware: Unspecified
OS: Unspecified
medium
medium
Target Milestone: ---
Assignee: Jiri Denemark
QA Contact: Virtualization Bugs
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2011-08-05 18:35 UTC by Dave Allan
Modified: 2018-07-18 14:13 UTC (History)
8 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2016-04-07 20:24:24 UTC


Attachments (Terms of Use)

Description Dave Allan 2011-08-05 18:35:43 UTC
Description of problem:
If I migrate a domain using:
virsh migrate --persistent --undefinesource --tunnelled --p2p --live --desturi 'qemu+ssh://root@hybrid0/system' spice

and restart the src libvirtd during the migration, the guest will remained defined but paused on the dst host.  Stopping the guest on the dst host causes it to vanish.  Migrating without the --persistent flag does not exhibit this behavior.

Version-Release number of selected component (if applicable):
libvirt-0.9.4-0rc1.2.el6.x86_64

How reproducible:
100%

Steps to Reproduce:
1. virsh migrate --persistent --undefinesource --tunnelled --p2p --live --desturi 'qemu+ssh://root@hybrid0/system' spice
2. during the migration, restart libvirtd on the src host
  
Actual results:
The migration is cancelled but the guest still exists, but is paused on the dst host.


Expected results:
The migration is cancelled and the guest is running on the src host and does not exist on the dst host.

Comment 1 Dave Allan 2011-08-05 18:39:05 UTC
One additional datapoint: the src libvirtd should be stopped and started after a few seconds, not restarted.  It seems to make a difference.

Comment 3 Jiri Denemark 2011-08-12 21:10:55 UTC
I was able to reproduce this bug with --p2p --tunnelled when source libvirtd crashed after calling Prepare on target but before actually starting to transmit any data. It doesn't reproduce with just --p2p. The difference is that after with --tunnelled, nothing happens after

virNetSocketReadWire:910 : End of file while reading data: Input/output error

In plain p2p mode, virConnectClose is called, which triggers qemuProcessAutoDestroyRun and the domain is removed.

Comment 8 Cole Robinson 2016-03-23 13:08:33 UTC
Has anyone verified this is still an issue? It hasn't been materially updated for over 4 years


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