Newer virtual machines may use nvram, and libvirt needs a special flag
to clean these up. Prior to this change (with libvirt-8.0.0-8.1.el9_0):
File "/usr/sbin/kojivmd", line 1072, in cleanupVM
vm.undefine()
File "/usr/lib64/python3.9/site-packages/libvirt.py", line 3178, in undefine
raise libvirtError('virDomainUndefine() failed')
libvirt.libvirtError: Requested operation is not valid: cannot undefine
domain with nvram
RHEL 7 first introduced the libvirt.VIR_DOMAIN_UNDEFINE_NVRAM flag in
libvirt-python-1.2.8-7.el7 (rhbz#1144284)
lookupByName() could raise libvirt.libvirtError for many different
reasons (libvirt connection problems, etc).
If this kojivmd host does not have this VM available
(VIR_ERR_NO_DOMAIN), and we should log the "VM not available" message
and skip taking the task.
If the error is something else, this is unexpected, and we should raise
it in the logs so the administrator can see it at non-debug log levels.
In RHEL 9 (qemu-img-6.2.0-11.el9_0.3), the "qemu-img create" command
requires the "-F" flag that specifies the second image's format.
qemu-img-1.5.3-175.el7_9.6 recognizes this -F flag also, so it should be
safe to use on all supported platforms.
This is a simple extention of `SCM.assert_allowed`
- `assert_allowed_by_policy` will set the default "use_common" to False which is different to the old behavior
- `channel`, `user_id`, `scratch` are passed in the `policy_data` with scminfo right now.
This is a prototype for this change, and there are some other solutions could be implemented too
- Use a scmpolicy plugin as `postSCMCheckout` callback, the pro is that we can do more checks after the source is initialized on builder, meanwhile, the con is that the source will be downloaded even it is denied by policy. It might be a potential risk?
- Do the scm check in hub's `make_task`, this looks straightforward, but may lack some builder's information
fixes: #2757
If there are multiple e-mailes separated by comma of space, they should
be split for correct usage of smtplib.sendmail.
Fixes: https://pagure.io/koji/issue/1240
The Koji data model has been updated to support multiple checksum types. These are listed in the
CHECKSUM_TYPES enum in koji/__init__.py, but are not available in kojikamid.py. This change
copies the Enum class and the CHECKSUM_TYPES enum into kojikamid at build time, and uses it when
validating the checksums of downloaded files.
kojikamid has the SCM class from daemon.py inserted into it at build time. This class
has been converted to use urllib from six.moves (for Python 3 compatibility), rather
than using urlparse. Update the import in kojikamid.py to be compatible with this change.
