os/bluestore: mark ops that can't tolerate ENOENT #12114

Merged
merged 2 commits into from Nov 28, 2016

Projects

None yet

2 participants

@liewegas
Member

If we get errors on these we screw up the transaction
data_bl decoding cursor; assert they don't happen. In
general, this should already be true for transactions
that the OSD is producing.

Signed-off-by: Sage Weil sage@redhat.com

@liewegas liewegas os/bluestore: mark ops that can't tolerate ENOENT
If we get errors on these we screw up the transaction
data_bl decoding cursor; assert they don't happen.  In
general, this should already be true for transactions
that the OSD is producing.

Signed-off-by: Sage Weil <sage@redhat.com>
b08df13
@liewegas liewegas added this to the kraken milestone Nov 23, 2016
@liewegas liewegas added the needs-qa label Nov 23, 2016
@liewegas liewegas os/filestore: blacklist same ENOENT failures as bluestore
FileStore doesn't care per se, but they indicate buggy calling code, and
we want to catch all bad callers!

Signed-off-by: Sage Weil <sage@redhat.com>
28c73c6
@athanatos
Member

Looks right to me assuming it passes tests.

@liewegas liewegas merged commit 6ecb44f into ceph:master Nov 28, 2016

3 checks passed

Signed-off-by all commits in this PR are signed
Details
Unmodifed Submodules submodules for project are unmodified
Details
default Build finished.
Details
@liewegas liewegas deleted the liewegas:wip-bluestore-failops branch Nov 28, 2016
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment