| Age | Commit message (Collapse) | Author |
|---|
|
We're running the CI server heavily loaded - bump these up again.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
|
|
test_device_repeated_add_remove(){
Test rapidly adding and removing the same device.
Caught bugs:
- found a race condition that leads to a kernel BUG in btree_key_cache.
test_add_label(){
Test setting device label when adding
Caught bugs:
- label was not set correctly; set to dev-0 label instead of new one.
test_add_compound_label(){
Test adding a device with a compond label, a '.' separated "path"
(e.g. ssd.sata)
Caught bugs:
- filesystem crashed with message "Invalid superblock section disk_groups".
|
|
This is needed or we get random mount failures when running a kernel
with all the filesystems enabled.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
Mostly bumping up some timeouts
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
This is the start of making .ktest tests programs that are executed
directly, instead of having them sourced and invoked by ktest.
- Kill require-lib
- list_tests, run_test, run_tests are now normal library functions,
instead of using code injection
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
Instead, if it's not specified there's a default 10 minute timeout - now
that we can specify per-test timeouts the global timeout isn't as
important.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
- longer timeout for tiering_buffered, drop_alloc
- migrate_fom_fs now works for xfs and ext4 (not btrfs)
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
Add a new test for verifying that KEY_TYPE_error keys are being created
correctly, and fix replace_replica - we need to mark the old device as
failed (or remove it) before kicking off rereplicate.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
|
|
This currently fails because metadata doesn't get copied to the
replacement device, although the data content of the filesystem
does get replicated during the data rereplicate step.
Signed-off-by: Chris Webb <chris@arachsys.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
