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: never send rados ack (only commit) #12451

Merged
merged 12 commits into from Dec 29, 2016

Conversation

Projects
None yet
3 participants
@liewegas
Member

liewegas commented Dec 12, 2016

No description provided.

@yuriw

This comment has been minimized.

Contributor

yuriw commented Dec 14, 2016

@liewegas pls rebase

`[yuriw@smithi020 build]$ git pull https://github.com/liewegas/ceph.git wip-osd-on-applied
remote: Counting objects: 20, done.
remote: Total 20 (delta 16), reused 16 (delta 16), pack-reused 3
Unpacking objects: 100% (20/20), done.
From https://github.com/liewegas/ceph

  • branch wip-osd-on-applied -> FETCH_HEAD
    Auto-merging src/osd/PrimaryLogPG.h
    CONFLICT (content): Merge conflict in src/osd/PrimaryLogPG.h
    Auto-merging src/osd/PrimaryLogPG.cc
    CONFLICT (content): Merge conflict in src/osd/PrimaryLogPG.cc
    Auto-merging src/osd/OSD.h
    Auto-merging src/osd/OSD.cc
    Automatic merge failed; fix conflicts and then commit the result.
    `
@liewegas

This comment has been minimized.

Member

liewegas commented Dec 20, 2016

@athanatos pushed a bunch of new patches removing replay support

@liewegas

This comment has been minimized.

Member

liewegas commented Dec 28, 2016

retest this please

liewegas added some commits Dec 10, 2016

osd/ReplicatedPG: kill hit_set_flushing
This was added before we started taking the ondisk lock in order to allow
reading a just-written hitset.  Since this is a rare path anyway, just
take the ondisk lock.

Signed-off-by: Sage Weil <sage@redhat.com>
osd: remove 'rlat' read latency metrics
These are measuring how long of a delay we see before ops are readable.
They are calculated in the on_applied path, which we'd like to remove,
and weren't reliably calculated anyway.

Signed-off-by: Sage Weil <sage@redhat.com>
osd/ReplicatedPG: do not send ACK separate from COMMIT
This never comes before COMMIT, making it useless to clients.  And it's
just unnecessary overhead.

Signed-off-by: Sage Weil <sage@redhat.com>
osd: we can discard "replay" ops
Signed-off-by: Sage Weil <sage@redhat.com>
osd/PG: do not queue ops for replay
We no longer send ACKs, so this can't happen anyway.

Signed-off-by: Sage Weil <sage@redhat.com>
osd/PG: do not go into replay state
Signed-off-by: Sage Weil <sage@redhat.com>
osd: remove replay_queue
Signed-off-by: Sage Weil <sage@redhat.com>
osd: simplify dup/resent logic
No need to special ack path.

Signed-off-by: Sage Weil <sage@redhat.com>
osd/PG: remove waiting_for_ack
No longer used.

Signed-off-by: Sage Weil <sage@redhat.com>
doc: remove crash_replay_interval mentions
Signed-off-by: Sage Weil <sage@redhat.com>
qa/workunits: remove crash_replay_interval tests
Signed-off-by: Sage Weil <sage@redhat.com>

@liewegas liewegas merged commit bc197d7 into ceph:master Dec 29, 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-osd-on-applied branch Dec 29, 2016

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