rbd: misc fixes #4113
rbd: misc fixes #4113
Conversation
|
FAIL: the output of run-make-check.sh on centos-7 for 016c411 is http://paste2.org/VXDB7VVj |
Sent from 
Otherwise it may cause conflicts when registering the lock. Signed-off-by: Zhiqiang Wang <zhiqiang.wang@intel.com>
Signed-off-by: Zhiqiang Wang <zhiqiang.wang@intel.com>
If length is clipped in aio_discard, the actual discarded length would be less than the passed in argument. Signed-off-by: Zhiqiang Wang <zhiqiang.wang@intel.com>
Should call the release function instead of deleting it to free librbd::RBD::AioCompletion and librbd::AioCompletion. Otherwise there is a potential memory leak. Signed-off-by: Zhiqiang Wang <zhiqiang.wang@intel.com>
Return earlier to avoid extra work of reading image data and omap if the detection fails or the destination image already exists. Signed-off-by: Zhiqiang Wang <zhiqiang.wang@intel.com>
|
FAIL: the output of run-make-check.sh on centos-7 for 6189a2e is http://paste2.org/pgBxscKB |
|
The failure of run-make-check.sh is weird. It fails at test/os/TestLFNIndex.cc. However the changes in this PR don't touch that part of the code. Look at the failure itself, it complains a just successfully created file can't be copied, saying that it doesn't exist. Anything wrong with the server running the script? Btw, I don't see this failure when running 'make check' locally. |
|
@athanatos, that sounds a bit like the bug you fixed a while back in
shared_cache... I think introduced with the new ObjectContext caching?
|
|
@wonzhq , I think the fails at test/os/TestLFNIndex.cc is not caused by your patch. My patch is also return fail because of test/os/TestLFNIndex.cc. You can repush your code directly. I think the test case in test/os/TestLFNIndex.cc would run fail sometimes. |
|
Thanks, @XinzeChi . I've already pushed the right code :) |
rbd: misc fixes Reviewed-by: Josh Durgin <jdurgin@redhat.com>
|
Thanks! Nice finds! |
No description provided.