Samuel Just's activity
From 11/21/2019 to 12/20/2019
12/20/2019
- 11:39 PM RADOS Bug #42328 (Resolved): osd/PrimaryLogPG.cc: 3962: ceph_abort_msg("out of order op")
- I can't check the original reports (logs have been removed), but assuming it's the same root cause PR #32382 5bb932c3...
- 01:31 AM RADOS Bug #42328: osd/PrimaryLogPG.cc: 3962: ceph_abort_msg("out of order op")
- I observed something similar on a ceph_test_rados teuthology run: sjust-2019-12-19_20:05:13-rados-wip-sjust-read-from...
- 11:37 PM Ceph Bug #43376 (Resolved): crimson::dmclock Assertion '__builtin_expect(__n < this->size(), true)' fa...
- Should be fixed with PR #32380 2c9542901532feafd569d92e9f67ccd2e1af3129
- 01:57 AM Ceph Bug #43376 (In Progress): crimson::dmclock Assertion '__builtin_expect(__n < this->size(), true)'...
- 01:57 AM Ceph Bug #43376: crimson::dmclock Assertion '__builtin_expect(__n < this->size(), true)' failed.
- Yeah, same bug.
- 11:37 PM RADOS Bug #43394 (Resolved): crimson::dmclock segv in crimson::IndIntruHeap
- Should be fixed with PR #32380 2c9542901532feafd569d92e9f67ccd2e1af3129
- 10:48 PM Ceph Revision 2c954290 (ceph): Merge pull request #32380 from athanatos/sjust/wip-dmclock-fix
- src/dmclock: bring in fixes for indirect_intrusive_heap
Reviewed-by: Josh Durgin <jdurgin@redhat.com> - 09:07 PM Ceph Revision bdc3ed25 (ceph): Merge pull request #32381 from athanatos/sjust/wip-read-from-replica-py2
- osd: propagate mlcod to replicas and fix problems with read from replica
Reviewed-by: Sage Weil <sage@redhat.com> - 09:07 PM Ceph Revision 5bb932c3 (ceph): Merge pull request #32382 from athanatos/sjust/wip-backoff-fix
- osd/scheduler/OpSchedulerItem: schedule backoffs as client ops
Reviewed-by: Sage Weil <sage@redhat.com> - 01:35 AM Ceph Revision 671c763a (ceph): PrimaryLogPG: drop replica obc cache upon repop
- Signed-off-by: Samuel Just <sjust@redhat.com>
- 01:35 AM Ceph Revision 69def143 (ceph): PG: drop replica reads from prior intervals
- Otherwise, they cause misdirected op warnings.
See comment for details.
Signed-off-by: Samuel Just <sjust@redhat.com> - 01:35 AM Ceph Revision 900bd37a (ceph): PrimaryLogPG: report ENOENT on missing clone on replica
- Can't check is_degraded_or_backfilling etc on replica.
Signed-off-by: Samuel Just <sjust@redhat.com> - 01:35 AM Ceph Revision a335a259 (ceph): osd/PGBackend: rename submit_transaction field roll_forward_to
- This field is actually just a lower bound on committed, client
visible log versions.
Signed-off-by: Samuel Just <sju... - 01:35 AM Ceph Revision 136d3baf (ceph): ReplicatedBackend: always rollforward on repop to version
- The primary always sets this field to be at_version, and we're
going to reuse this field to mean min_last_complete_on... - 01:35 AM Ceph Revision 6f12bf27 (ceph): ReplicatedBackend,MOSDRepOp: replace pg_roll_forward_to with mlcod
- New clients will ignore the field on header.version <= 2. New
primaries will send the old variant if there are peers... - 01:35 AM Ceph Revision a8ad5090 (ceph): osd/: maintain PeeringState::min_last_complete_ondisk on replica
- Signed-off-by: Samuel Just <sjust@redhat.com>
- 01:35 AM Ceph Revision 044fa545 (ceph): osd/: bounce writes on objects written since mlcod back to primary
- If there are any writes since mlcod on an object with the same head,
our ondisk state may be unstable or may reflect ... - 01:29 AM Ceph Revision bb848ed6 (ceph): osd/scheduler/OpSchedulerItem: schedule backoffs as client ops
- The backoff machinery relies on backoffs being ordered the same
way as client ops.
Signed-off-by: Samuel Just <sjust...
12/19/2019
- 10:55 PM Ceph Revision 7b2cba66 (ceph): src/dmclock: bring in fixes for indirect_intrusive_heap
- Signed-off-by: Samuel Just <sjust@redhat.com>
12/18/2019
- 11:46 PM Ceph Bug #43376: crimson::dmclock Assertion '__builtin_expect(__n < this->size(), true)' failed.
- [ RUN ] IndIntruHeap.remove_greatest
/usr/include/c++/8/bits/stl_vector.h:932: std::vector<_Tp, _Alloc>::refere... - 10:05 PM Ceph Bug #43376 (Resolved): crimson::dmclock Assertion '__builtin_expect(__n < this->size(), true)' fa...
- sage-2019-12-16_15:01:24-rados-wip-sage3-testing-2019-12-13-1721-distro-basic-smithi
It's a bug with dmclock/suppo...
12/03/2019
- 05:35 AM Ceph Revision 030c0570 (ceph): vstart: add --debug for crimson if debug is enabled
- Signed-off-by: Samuel Just <sjust@redhat.com>
- 05:35 AM Ceph Revision 260f23b7 (ceph): crimson/common/errorator.h: add finally method for resource release
- This way we can add things like resource handlers which are otherwise
transparent to passing values and errors.
Sign... - 05:35 AM Ceph Revision 83eba36c (ceph): osd/: factor OSD::init_op_flags into seperate class
- We'll want to reuse this in crimson, extract the logic
for setting flags from MOSDOp into osd_types.h.
Signed-off-by... - 05:35 AM Ceph Revision fc8f8855 (ceph): common/: introduce intrusive_lru
- Signed-off-by: Samuel Just <sjust@redhat.com>
- 05:35 AM Ceph Revision 2472b376 (ceph): src/osd: factor out RWState, move SnapSetContext, ObjectState, RWState
- We'll want to reuse this logic in crimson, but the actual operation
wake/wait mechanics will be different.
Signed-of... - 05:35 AM Ceph Revision f35c90c7 (ceph): crimson: add ObjectContext/Registry
- ObjectContext will hold the ephemeral state and lock status
for an object as in classic osd.
Signed-off-by: Samuel J... - 05:35 AM Ceph Revision fd803498 (ceph): crimson/osd/pg_backend: factor out load_object_state/snapset
- Signed-off-by: Samuel Just <sjust@redhat.com>
- 05:35 AM Ceph Revision ffb66d4a (ceph): crimson/osd: use ObjectContext and take obc locks
- Signed-off-by: Samuel Just <sjust@redhat.com>
- 05:35 AM Ceph Revision e5a0499b (ceph): crimson/osd: convert PG::wait_for_active to a blocking_future
- Signed-off-by: Samuel Just <sjust@redhat.com>
Also available in: Atom