Bug 971485
| Summary: | Live migration fails if "vnc_listen" set in /etc/libvirt/qemu.conf | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 6 | Reporter: | Roman <rommer> |
| Component: | libvirt | Assignee: | Michal Privoznik <mprivozn> |
| Status: | CLOSED ERRATA | QA Contact: | Virtualization Bugs <virt-bugs> |
| Severity: | medium | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | 6.4 | CC: | acathrow, dallan, dyuan, lcui, pasquale.demichele, ydu, zpeng |
| Target Milestone: | rc | ||
| Target Release: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | libvirt-0.10.2-20.el6 | Doc Type: | Bug Fix |
| Doc Text: |
Cause:
If user specified no vnc listen address in their domain XML, the one from qemu.conf is used. However, uppon migrating, there was no difference between cases where the listen address was set by user in XML directly or copied from qemu.conf.
Consequence:
A domain couldn't be migrated.
Fix:
If the listen address is copied from qemu.conf, it is not transferred to the destination (which probably can't bind() to the old listen address anyway).
Result:
A domain can be migrated without any error.
|
Story Points: | --- |
| Clone Of: | Environment: | ||
| Last Closed: | 2013-11-21 09:02:25 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: | |||
|
Description
Roman
2013-06-06 16:01:38 UTC
Patch that breaks live migration: http://libvirt.org/git/?p=libvirt.git;a=commit;h=d7f9d827531bc843b7c5aa9d3e8c08738a1de248 Roman, thanks for the bisect; that makes sense. Michal, can you see what can be done about this? Simple workaround using libvirt hooks: # cat /etc/libvirt/hooks/qemu-filter.xsl <?xml version="1.0"?> <xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"> <xsl:output omit-xml-declaration="yes"/> <xsl:template match="/domain/devices/graphics/listen"/> <xsl:template match="/domain/devices/graphics/@listen"/> <xsl:template match="@*|node()"> <xsl:copy> <xsl:apply-templates select="@*|node()"/> </xsl:copy> </xsl:template> </xsl:stylesheet> # cat /etc/libvirt/hooks/qemu #!/bin/bash [ "$2" = "migrate" ] && xsltproc /etc/libvirt/hooks/qemu-filter.xsl - exit 0 Patch proposed upstream: https://www.redhat.com/archives/libvir-list/2013-June/msg00456.html I've just pushed the patch upstream:
commit 6546017c50c104d0b9867137b64ab1f4a312e436
Author: Michal Privoznik <mprivozn>
AuthorDate: Mon Jun 10 17:51:46 2013 +0200
Commit: Michal Privoznik <mprivozn>
CommitDate: Tue Jun 11 14:11:46 2013 +0200
qemu_migrate: Dispose listen address if set from config
https://bugzilla.redhat.com/show_bug.cgi?id=971485
As of d7f9d827531bc843b7c5aa9d3e8c08738a1de248 we copy the listen
address from the qemu.conf config file in case none has been provided
via XML. But later, when migrating, we should not include such listen
address in the migratable XML as it is something autogenerated, not
requested by user. Moreover, the binding to the listen address will
likely fail, unless the address is '0.0.0.0' or its IPv6 equivalent.
This patch introduces a new boolean attribute to virDomainGraphicsListenDef
to distinguish autofilled listen addresses. However, we must keep the
attribute over libvirtd restarts, so it must be kept within status XML.
v1.0.6-61-g6546017
# rpm -q libvirt libvirt-0.10.2-20.el6.x86_64 Following the steps in comment0, when start migration, libvirtd on dest host will crash, which should cause by bug 984793. 1. Set vnc_listen="10.66.106.30" in /etc/libvirt/qemu.conf
2. Start a guest
# virsh dumpxml spice
...
<graphics type='vnc' port='5900' autoport='yes' listen='10.66.106.30'>
<listen type='address' address='10.66.106.30'/>
</graphics>
...
3. Do live migration
3.1 With libvirt-0.10.2-18.el6_4.5, when do migration:
# virsh migrate --timeout 300 --live --verbose spice qemu+tcp://10.66.106.31/system
error: internal error process exited while connecting to monitor: char device redirected to /dev/pts/2
inet_listen_opts: bind(ipv4,10.66.106.30,5900): Cannot assign requested address
inet_listen_opts: FAILED
Failed to start VNC server on `10.66.106.30:0'
3.2. With libvirt-0.10.2-21.el6, when do migration:
# virsh migrate --timeout 300 --live --verbose spice qemu+tcp://10.66.106.31/system
Migration: [100 %]
and on the dest host:
# virsh list
Id Name State
----------------------------------------------------
1 spice running
as vnc_listen="10.66.106.31 on the dest host qemu.conf, the guest listen address also changed
# virsh dumpxml spice
...
<graphics type='vnc' port='5900' autoport='yes' listen='10.66.106.31'>
<listen type='address' address='10.66.106.31'/>
</graphics>
...
So move this bug to VERIFIED.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHBA-2013-1581.html (In reply to errata-xmlrpc from comment #12) > Since the problem described in this bug report should be > resolved in a recent advisory, it has been closed with a > resolution of ERRATA. > > For information on the advisory, and where to find the updated > files, follow the link below. > > If the solution does not work for you, open a new bug report. > > http://rhn.redhat.com/errata/RHBA-2013-1581.html Hi all. I have the same issue with this packages: $ rpm -qa | grep libvirt libvirt-python-0.10.2-29.el6_5.3.x86_64 libvirt-client-0.10.2-29.el6_5.3.x86_64 libvirt-0.10.2-29.el6_5.3.x86_64 OS version is the same, CentOS 6.4. The cloud system is OpenNebula (4.4) and the migration fails also when the listen_vnc (different from 0.0.0.0) is setted directly in the template. |