mds/MDBalancer: add an arg to limit depth when dump loads for dirfrags
Reviewed-by: Venky Shankar <vshankar@redhat.com>
Reviewed-by: Ramana Raja <rraja@redhat.com>
mgr/stats: change in structure of perf_stats o/p
Reviewed-by: Venky Shankar <vshankar@redhat.com>
Reviewed-by: Anthony D'Atri <anthony.datri@gmail.com>
Reviewed-by: Jos Collin <jcollin@redhat.com>
mds/client: fail the request if the peer MDS doesn't support getvxattr op
Reviewed-by: Venky Shankar <vshankar@redhat.com>
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Milind Changire <mchangir@redhat.com>
test_client_metrics_and_metadataand other tests has been
updated as earlier it was checking according to the old structure
of perf stats o/p, which has been changed in this PR.
Fixes: https://tracker.ceph.com/issues/56162
Signed-off-by: Neeraj Pratap Singh <neesingh@redhat.com>
Before unmounting check if the client has been evicted and, if so, run
"umount -f -l" for the mount point of the client and cleanup the mount
right after it.
Attempting to unmount, cleanup or operate in any way over mount point
of a evicted client will hang the operation (and thereby our Python
code too). Lazy-force unmount prevents such hangs for our Python code
and also frees the mount point.
This commit also adds code to gather session info for kernel mounts
after mounting is successful. This is a necessity since network address
of session is needed to check if it is blocked by Ceph cluster.
Fixes: https://tracker.ceph.com/issues/56476
Signed-off-by: Rishabh Dave <ridave@redhat.com>
For new OPs the old ceph may will crash the MDSs when receiving
unknow OPs. As a workaround the new clients will check the sessions
feature bits to decide to send the new OPs or not.
Fixes: https://tracker.ceph.com/issues/56529
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Old cephs, such as nautilus, will use "blacklist" instead of "blocklist".
Fixes: https://tracker.ceph.com/issues/56529
Signed-off-by: Xiubo Li <xiubli@redhat.com>
For old cephs, such as the nautilus, there is no --pg_num_min option.
Fixes: https://tracker.ceph.com/issues/56529
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Internal directories: '_nogroup', '_index', '_legacy', '_deleting'
1. Internal directories should be filtered in 'subvolmegroup ls' command.
2. Internal directories should not be accepted as a group name.
Fixes: https://tracker.ceph.com/issues/55762
Signed-off-by: Nikhilkumar Shelke <nshelke@redhat.com>
Validate subvolume discover on upgrade from
legacy subvolume to v1. The handcrafted
`.meta' file on legacy subvolume root should
not be used for any subvolume apis like getpath,
authorize.
Signed-off-by: Kotresh HR <khiremat@redhat.com>
Remove setup_test_env() because removing this methods removes an extra
layer of abstraction which makes tests more readable. Rename
_create_client_and_keyring_file() to just _create_client() and reverse
the order of parameters in remount_with_new_client() and set default
value of cephfs_mntpt to '/'.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
test_single_path_rootsquash instead of using helper methods duplicates
the code from those methods. This commit fixes that and also upgrade
this method since caps_helper was upgraded in previous commits.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
Upgrade test_admin.FsAuthorize to repair its compatibility with
caps_helper.py since caps_hepler.py has been heavily modified in
previous commits.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
caps_helper.py since caps_helper.py has been heavily modified in
previous commits.
setup_fs_contents() is being deleted since it is not used anymore.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
The intention is to make logs contain some information of what's being
done by caps_helper.py regardless of which test/method is calling it.
This should make logs more understandable and also add hints for
future debugging.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
With this commit, the design of run_mon_cap_tests() is now aligned with
rest of CapTester -- it's not meant to be inherited by test class
anymore and it is to be called by using CapTester instance.
This commit also changes working of this method. Now instead of
obtaining FS names from Python objects representing FSs, it obtains
those names from MON cap. This removes the need to pass the FS objects
around.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
Let's change the file data to include the file path, the CephFS name
and the host FS and CephFS mountpoint so that the bugs where, let's say,
"cephfs2" is mounted instead of "cephfs1" (where obviously both the
CephFSs lie on the same Ceph cluster) is caught by the tests due to
uniqueness of every test file's content.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
Simplify methods in CapsTester by adding a test set, which will be a
list of tuples. The first element in tuple will be the mount object,
the second will be the test file path and the third will be the test
file data. Thus instead of having three independent class variables
that are always used together now we have list of test sets making
management of multiple test sets simpler.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
Modify caps_helper.py such that calling code does't have to store
returned values just to pass those values again to a method in
caps_helper.py. This is a common pattern where write_test_files()
return 3 values and all 3 passed as it is to run_cap_tests().
The easy way to do it is to upgrade caps_helper so that it can be used
an object and not just as a class supplying helper methods. The return
values will be stored by the object internally and thus resued. In case
of testing multiple FSs inside a single test method, we'll now need
multiple instances of this class to keep those values separate.
And since CapsHelper is not just a class supplying helper methods
anymore, it's being renamed to CapTester.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
Concerned method: caps_helper.CapsHelper.write_test_file()
When CephFS mountpoint is changed inside a single test method, the test
files created at root are neither accessible nor useful. Therefore, add
an option to create the test files in the directory passed by the user.
This also increases general flexibility of the concerned method.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
Instead of relying on value of a mutable variable, actually check if the
CephFS is mounted on the system. This will prevent bugs due to stale and
incorrect values.
Fixes: https://tracker.ceph.com/issues/54283
Signed-off-by: Rishabh Dave <ridave@redhat.com>
Method authorize currently accepts caps as a list that should be as
follows -
(path1, perm2, path2, perm2)
Modify this method to also accepts caps arranged in following format -
((path1, perm1), (path2, perm2), (path3, perm3))
The latter format is pairs the path and the permission to be set on path
together. This makes the association more apparent and thus it less
prone to human error.
Signed-off-by: Rishabh Dave <ridave@redhat.com>
A new path is obtained every iteration which is appended to command
arguments. After the command has been executed successfully, delete the
the path from command's arguments so that the command's arguments don't
have two paths in next iteration and it works as expected.
Fixes: https://tracker.ceph.com/issues/56416
Signed-off-by: Rishabh Dave <ridave@redhat.com>
