Description of problem:
libvirts use 3 URIs for migration:
1) source libvirtd URI
2) destination libvirtd URI
3) destination QEMU URI.
And if the destination QEMU URI is not specified by user, libvirt will try to execute "hostname()" on destination host to guess the hostname, and uses it to
form QEMU URI, (e.g. tcp://10.66.70.98).
So, If destination host is mis-configured, it will return a illed hostname which would not be able to address by source host, as a result, the migration will fail.
The problem here is that qemu discard the errors for migration failing (actually all of the errors related to getaddrinfo(3)/connect(2) failing?), so, libvirt can not get the exact error information for dignose log, an error like "An undefined error has occured" obviously is not what the user want to see.
To get more info, see:
http://post-office.corp.redhat.com/mailman/private/virt-devel/2011-January/msg00239.html
Version-Release number of selected component (if applicable):
How reproducible:
Steps to Reproduce:
1.
2.
3.
Actual results:
Expected results:
Additional info:
Hi,Juan/Osier
would you please tell me is this issue duplicate Bug 584077 - QMP: Improve migrate command error info? seems bz584077 is key reason, libvirt call qmp.
Comment 7Markus Armbruster
2011-07-28 14:52:24 UTC
Yes, this is the same as bug 584077.
*** This bug has been marked as a duplicate of bug 584077 ***