fe2o3_o3db_sync/tests/chunk_set.rs
14 operations, oldest first
created by r1870400018:69929, which is this file's identity for as long as the history lasts, whatever it is later renamed to
read the file · who wrote it · compare two states
An operation carries no clock of its own. Each row is therefore dated by the first mark written after it by the same replica, which is one copy of the repository, and that mark is where the command that wrote the operation ended.
+ signed and verified, ? signed by a key this forge does not carry, - unsigned
| Operation | Author | What | When | |
|---|---|---|---|---|
| + | r1870400018:69929 | replica 1870400018 | created fe2o3_o3db_sync/tests/chunk_set.rs | 42 days ago under "Merge origin/main into dozone-chunkset" |
| + | r1870400018:69930 | replica 1870400018 | inserted 33.6 KiB | 42 days ago under "Merge origin/main into dozone-chunkset" |
| + | r1870400018:70265 | replica 1870400018 | inserted 4.8 KiB | 42 days ago under "fe2o3_o3db_sync: a stamp floor and one tie order -- a write after a restart outranks every record on disk even after the clock steps back, and two records of a key with one stamp are ordered by place in every path, so a key reads the same value live and after a restart (Ozone run 1) Replay raises the process's stamp clock past every stamp it reads (api::raise_stamp_floor in both bot_initgc replay paths) before start returns, and every record, tombstones included, is stamped through api::write_stamp (pending.rs own-set retire, data/core.rs zero-time record); KeyVal::stamp_time_now is gone. data::cache::supersedes orders by (time, fnum, offset) and is shared by Cache::insert, replay and bot_scan." |
| + | r1870400018:70406 | replica 1870400018 | inserted 6 bytes | 42 days ago under "fe2o3_o3db_sync: retire on supersession, hold until final -- a store retires the chunk set of the record its durable head displaced, as the cache bot answers it, instead of reading the key back, so overlapping overwrites and deletes leave no live orphan; a chunk set is held from the draw of its id until its head's outcome is final, so the orphan sweep never takes a set in flight (Ozone run 2) The cache bot that enters a ticketed head answers HeadInserted(cind, Displaced): Nothing, Itself (a newer record was already there: the store retires its own set), or Record { meta, prior } with Prior::Head(bytes), HeadUnread (a replayed head, matched against the value read before the store began by metadata) or NotHead. Heads never enter the write tracker; Cache::insert moves the old value's bytes out on supersession, so a newer valueless record no longer leaves a stale value under a new location. The writer releases a hold only when the syncer check or the Data append fails; the cache bot releases it after an insert, confirmed or not, and keeps it on an insert error. The forcing path (test support) is unticketed and retires nothing. A delete is a store with no chunks whose tombstone is the head. Retire tombstones are stamped by api::retire_stamp, the ordered clock without the stale-stamp test hook." |
| + | r1870400018:70407 | replica 1870400018 | inserted 6.0 KiB | 42 days ago under "fe2o3_o3db_sync: retire on supersession, hold until final -- a store retires the chunk set of the record its durable head displaced, as the cache bot answers it, instead of reading the key back, so overlapping overwrites and deletes leave no live orphan; a chunk set is held from the draw of its id until its head's outcome is final, so the orphan sweep never takes a set in flight (Ozone run 2) The cache bot that enters a ticketed head answers HeadInserted(cind, Displaced): Nothing, Itself (a newer record was already there: the store retires its own set), or Record { meta, prior } with Prior::Head(bytes), HeadUnread (a replayed head, matched against the value read before the store began by metadata) or NotHead. Heads never enter the write tracker; Cache::insert moves the old value's bytes out on supersession, so a newer valueless record no longer leaves a stale value under a new location. The writer releases a hold only when the syncer check or the Data append fails; the cache bot releases it after an insert, confirmed or not, and keeps it on an insert error. The forcing path (test support) is unticketed and retires nothing. A delete is a store with no chunks whose tombstone is the head. Retire tombstones are stamped by api::retire_stamp, the ordered clock without the stale-stamp test hook." |
| + | r1870400018:70584 | replica 1870400018 | replaced 6 bytes with 7 bytes | 42 days ago under "fe2o3_o3db_sync: a set retires only against a confirmed winner, a pinned-heads gauge, and a sweep that reads no clock -- a record beaten or displaced only by an unconfirmed one is neither old nor an orphan, and every set let go during a sweep is kept by the sweep that watched it (Ozone R2-1, R2-5, R1-1)" |
| + | r1870400018:70589 | replica 1870400018 | inserted 6 bytes | 42 days ago under "fe2o3_o3db_sync: a set retires only against a confirmed winner, a pinned-heads gauge, and a sweep that reads no clock -- a record beaten or displaced only by an unconfirmed one is neither old nor an orphan, and every set let go during a sweep is kept by the sweep that watched it (Ozone R2-1, R2-5, R1-1)" |
| + | r1870400018:71172 | replica 1870400018 | replaced 11 bytes with 140 bytes | 40 days ago under "fe2o3_o3db_sync: a failed cut back whose fault record cannot be written is held by the refusal marker alone, which carries its cut, and clearing the marker folds the cut into the record so the next open cuts the pair back -- a test now pins that path of the round-2 marker; and the torn-append error no longer carries a run of spaces from a string continued without its backslash (Ozone APPENDSEEK, rebase onto round 2)" |
| + | r1870400018:71173 | replica 1870400018 | inserted 1 byte | 40 days ago under "fe2o3_o3db_sync: a failed cut back whose fault record cannot be written is held by the refusal marker alone, which carries its cut, and clearing the marker folds the cut into the record so the next open cuts the pair back -- a test now pins that path of the round-2 marker; and the torn-append error no longer carries a run of spaces from a string continued without its backslash (Ozone APPENDSEEK, rebase onto round 2)" |
| + | r1870400018:71174 | replica 1870400018 | replaced 59 bytes with 184 bytes | 40 days ago under "fe2o3_o3db_sync: a failed cut back whose fault record cannot be written is held by the refusal marker alone, which carries its cut, and clearing the marker folds the cut into the record so the next open cuts the pair back -- a test now pins that path of the round-2 marker; and the torn-append error no longer carries a run of spaces from a string continued without its backslash (Ozone APPENDSEEK, rebase onto round 2)" |
| + | r1870400018:71175 | replica 1870400018 | replaced 5 bytes with 14 bytes | 40 days ago under "fe2o3_o3db_sync: a failed cut back whose fault record cannot be written is held by the refusal marker alone, which carries its cut, and clearing the marker folds the cut into the record so the next open cuts the pair back -- a test now pins that path of the round-2 marker; and the torn-append error no longer carries a run of spaces from a string continued without its backslash (Ozone APPENDSEEK, rebase onto round 2)" |
| + | r1870400018:71177 | replica 1870400018 | inserted 39 bytes | 40 days ago under "fe2o3_o3db_sync: a failed cut back whose fault record cannot be written is held by the refusal marker alone, which carries its cut, and clearing the marker folds the cut into the record so the next open cuts the pair back -- a test now pins that path of the round-2 marker; and the torn-append error no longer carries a run of spaces from a string continued without its backslash (Ozone APPENDSEEK, rebase onto round 2)" |
| + | r1870400018:71178 | replica 1870400018 | inserted 55 bytes | 40 days ago under "fe2o3_o3db_sync: a failed cut back whose fault record cannot be written is held by the refusal marker alone, which carries its cut, and clearing the marker folds the cut into the record so the next open cuts the pair back -- a test now pins that path of the round-2 marker; and the torn-append error no longer carries a run of spaces from a string continued without its backslash (Ozone APPENDSEEK, rebase onto round 2)" |
| + | r1870400018:71180 | replica 1870400018 | replaced 19 bytes with 14 bytes | 40 days ago under "fe2o3_o3db_sync: a failed cut back whose fault record cannot be written is held by the refusal marker alone, which carries its cut, and clearing the marker folds the cut into the record so the next open cuts the pair back -- a test now pins that path of the round-2 marker; and the torn-append error no longer carries a run of spaces from a string continued without its backslash (Ozone APPENDSEEK, rebase onto round 2)" |