clusterAdditionalInfo map is holding a localClusterState
for checking ceph cluster supports resize and subvolumegroup
is created or not, currently we are checking if the key is present
in a map and localClusterStatelocalClusterState.resizeSupported
is set to false to call ceph fs subvolume resize to check command is
supported or not, if a structure is initialized all its members
are set to default value. so we will never going to check the
ceph fs subvolume resize command is supported in backend or not, we are
always using ceph fs subvolume create to resize subvolume. in some
ceph version ceph fs subvolume create wont work to resize a subvolume.
This commit changes the resizeSupported from bool to *bool for
proper handling of this scenario.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
(cherry picked from commit a35a835e9c)
if the cephfs mountpoint is not mounted the umount
call will fail with both stderr and err,
an error (exit status 32) and stdError (umount: /var/lib/kubelet/plugins/kubernetes.io/csi/pv/pvc-e622cd2b-0b87-4f49-a709-b89664ca6ec4/globalmount: not mounted.
checking stderr is sufficient to find out the mountpoint
is not mounted.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
sometimes, cephfs returns pending as clone
status if we request for the subvolume clone.
cephcsi need to consider the pending state
when checking the clone status and it should
return abort error message if the clone is in
pending state or in-progress state.
Co-authored-by: Yug <yuggupta27@gmail.com>
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
currently in someplaces we are checking the stderror
for CLI errors and insome places we are checking for
just error. This commit fixes the issue by checking
the stderror for actual CLI errors.
Co-authored-by: Yug <yuggupta27@gmail.com>
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
Previously the purgeVolume error was ignored due to wrong error variable
check in the createVolume. With this change it checks on the proper error.
Signed-off-by: Humble Chirammal <hchiramm@redhat.com>
(cherry picked from commit 9dee064b77)
return a proper error message to the user when
the subvolume has the snapshots and it cannot
be removed until the snapshots on the subvolume
have to be deleted.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
(cherry picked from commit 89b326c896)
When passing
fuseMountOptions: debug
in the StorageClass, the mount options passed on the ceph-fuse
commandline result in "-o nonempty ,debug". The additional space before
the ",debug" causes the mount command to fail.
Fixes: 1485
Signed-off-by: Niels de Vos <ndevos@redhat.com>
(cherry picked from commit 6f2f972f5b)
Get subvolume features using subvolume info and update the
volume options with Features parameter.
Signed-off-by: Yug <yuggupta27@gmail.com>
(cherry picked from commit 0682f0e500)
The subvolume features consists the list of features, which if includes
"snapshot-autoprotect", result in query based approach to
detecting the need for protect/unprotect.
If "snapshot-autoprotect" feature is present, The UnprotectSnapshot
call should be treated as a no-op
Signed-off-by: Yug <yuggupta27@gmail.com>
(cherry picked from commit 9e959c45a3)
The subvolume features consists the list of features, which if includes
"snapshot-autoprotect", result in query based approach to
detecting the need for protect/unprotect.
If "snapshot-autoprotect" feature is present, The ProtectSnapshot
call should be treated as a no-op
Signed-off-by: Yug <yuggupta27@gmail.com>
(cherry picked from commit 43a4c100ec)
Use subvolume info to fetch the subvolume path.
If `subvolume info` command is not available,
use `getpath` command instead.
Signed-off-by: Yug <yuggupta27@gmail.com>
(cherry picked from commit 365eb58ff7)
Add validateSnapshotRetention() to verify if subvolume has
`snapshot-retention` feature.
Signed-off-by: Yug <yuggupta27@gmail.com>
(cherry picked from commit 59fc4aa00f)
Snapshots can be retained even after subvolume deletion in
Ceph 14.2.12. Adding support for the same in ceph-csi.
Signed-off-by: Yug <yuggupta27@gmail.com>
(cherry picked from commit 6351358752)
It doesnot make sense to allow the creation of empty
volumes with readonly access, this commit allows the
creation of volume which is having readonly capabilities
only if the content source is set for the volume.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
(cherry picked from commit 860b65c446)
we should not return the CLI errors in GRPC errors
we need to return proper readable error messages
to the user for better understanding and better
debugging.
updates #1242
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
(cherry picked from commit 0bd45a5eb1)
currently the lock is not released which is
taken on the request name. this is causing issues
when the subvolume is requested for delete.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
(cherry picked from commit 787ba69089)
execCommandErr returns both error and stderror
message. checking strings.HasPrefix is not helpful
as the stderr will be the first string. its good
to do string comparison and find out that error
is volume not found error.
Signed-off-by: Madhu Rajanna <madhupr007@gmail.com>
(cherry picked from commit 31c7c95fdb)
