zvmsdk is a library that enables zvm cloud management.
This patch switch from calling xcat client directly to importing
zvmsdk to call zvmsdk API, to manage zvm host and virtual machines.
Change-Id: I74406a618a4a8b2d80a905c56621c1acd46f1c08
Currently, nova zvm virt driver is in nova namespace. This result
in some troubles when generating config file, generating docs and
packaging.
All changes in this commit are trying to move nova zvm virt driver
code into new namespace - nova_zvm .
Change-Id: I5251069dfd24ff4e337e9308439415482eb2234c
When boot-from-volume fails, its instance_fcp_map should be updated
correctly. Unfortunately, current code is wrong and will lead to an
error raised.
Change-Id: I7aa65d551b435e0600a40e652478a1fb775d5ca1
Right now, if delete userid failed due to image lock or device
lock, zvm driver has a looping call to check lock status with 10
minutes timeout.
Normaly, the image or device was locked by disk related actions
which could not be unlocked in short time. So turn to call
image_unlock_dm try to unlock it. Then try to delete the vm again.
Change-Id: I848d042231757d8673700e3db2b46f8fc9757596
Now the root_disk_units just has disk size in it. It doesn't have
units. So if customer specify a dismatch disk type, the driver
will not detect it until the create disk prompt the space is not
enough. This patch add units to root_disk_units, so it can give
customer clearly advise if error happens.
Change-Id: I1a31aa4bf77b774092fd04cc7174599252757538
zvm_default_admin_userid will be added and user is able to
log on cloud created user through this ID.
the old zvm_user_default_password will be removed in the near future.
part of blueprint:add_zvm_default_admin_userid
Change-Id: I8938669af3103df2e6f51cee1de458a6be339154
REST API layer might return 503 error, this means server is
temply not available, instead just fail, we can retry the work
as our REST calls are sync call, so no harm to retry it
as we can't retry forever, add a retry max count so that at
last we will still fail out.
Change-Id: I1e484265d241c87d2f0060f0aaef19cb6314ed4f
An instance object should have the os and version information so
that a volume can be attached to it when it's inactive.
Change-Id: I232f743a446dbfcaaf3794df02c084515bdd45fe
When boot a system with a NIC which is different from the master system,
NIC qeth device will not be initialized when system re-ipl.
Change-Id: I3afd2b7bef51db113212fde039cb32e902d6f81a
remove prefix('\u') of chinese unicode when snapshotting
or booting an instance
the image name will store as the format without prefix,like:
'\u6d4b\u8bd5\u7528\u4f8b' --> '6d4b8bd575284f8b'
add a function in util.py for code reuse
add a test for this function
Change-Id: I1bfc6ace2b74e4cfab3a6c6a36c31629e4290bf4
The hostname of the instance is not set correct when booting from
volume. It's caused by redundant data in network_metadata in
instance_metadata, so the data will be cleaned.
Change-Id: Id6c847cae4442fa70f2de212c22968973f139601
We use one WWPN to do boot-from-volume, but SVC driver always
returns a WWPN list from Newton. So we have to remove the
redundant ones.
Change-Id: I6c43eb8f057c79b3f9f75a8532608e5f42e924b1
In current spawn, at the beginning of spawn the root device is
set to /dev/sda automatically, that is not correct for a normal
spawn vm which should be /dev/dasda, so change to update the root
device in earlier stage of spawn, this change has no effect to
the function.
Change-Id: I7d83bad7844d56cc4bcbfd88b8f947c1a6d99ec1
Previously, we use different commands to set admin password for
different linux distros. However, we found this can be done by
chpasswd command on all distros, so change to use it.
Change-Id: I0cbbb4644e12fe8fb773fefbfe112b7bc2b0444d
add comments to describe the reason of adding semaphore and simply
introduce the problem to solve for future reference
Change-Id: I5dd03f5d47264df8a1bb8c3ed21b79407f867210
use a semaphore from eventlet to syschronize the image shared
between instances while booting an instance
Change-Id: I115e4f0c1686aa999a549d918df267c0459766a3
When boot from volume, image_meta has no id or other properties
it will cause the error like bug 1657 described. Submit this patch
to solve it.
Change-Id: Ie513ad7003b391d7012df946c3b660ae127aca1c
When make deployment, openstack will punch the policy files to set selinux policy
to allow socket communication to VMs.
Should work with zhcp's patch which include the policy file
Change-Id: Ia687b2f8341f541437e0e8f8103d8d07834f8075
zvm_vmrelocate_immediate = yes will give benefit to total migration
time but will increase the quiesence time, so we need set it to
no to get more benefit to end user.
Change-Id: Ib2091067e297a15a3409bf2dd6f9aa8a06ba7df4
This patch is for VM migrated from liberty, some of them can't setup IUCV,
so use SSH to make communication. We should keep the status as original VM.
Change-Id: I052ab748f61b1c05dbe8780e368b2d62aadaee60
we should raise nova exception so during build, nova
compute know it's a invalid image and stop retry
Change-Id: Id106d2c70fd12ffbd924ab20cd4022a3a9899f17
Print the details of the error which raised in processing of
driver._detach_volume_from_instance, instead of just return a
simple message.
Change-Id: I7b4353847e024a8de96ec6810c24dd2e1e51bb3c
Huang Rui