New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

osd: fix stat sum update of recovery pushing #13328

Merged
merged 1 commit into from Feb 14, 2017

Conversation

Projects
None yet
3 participants
@wonzhq
Contributor

wonzhq commented Feb 9, 2017

Use the stat field of PushInfo to do the stat update, which is
more accurate, especially in the case of sparse file. Also,
remove the stat sum update in on_peer_recover, which is done in
on_global_recover.

Signed-off-by: Zhiqiang Wang zhiqiang@xsky.com

Zhiqiang Wang
osd: fix stat sum update of recovery pushing
Use the stat field of PushInfo to do the stat update, which is
more accurate, especially in the case of sparse file. Also,
remove the stat sum update in on_peer_recover, which is done in
on_global_recover.

Signed-off-by: Zhiqiang Wang <zhiqiang@xsky.com>
get_parent()->release_locks(pi->lock_manager);
object_stat_sum_t stat = pi->stat;

This comment has been minimized.

@tchaikov

tchaikov Feb 9, 2017

Contributor

i'd suggest move this into the if (pushing[soid].empty()) { branch. and better off declare it as a reference. also where *_recovered field are set? we don't set them in ReplicatedBackend::prep_push().

This comment has been minimized.

@wonzhq

wonzhq Feb 9, 2017

Contributor

pi is erased from pushing and set to NULL before the if branch. See the two lines following.
The *_recovered fields are set in build_push_op.

@tchaikov tchaikov added the core label Feb 9, 2017

@yuriw yuriw merged commit 58a0c32 into ceph:master Feb 14, 2017

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
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment