Botond Dénes
Botond Dénes
> Should we make it true when tombstone_gc = repair? One might also want to enable this for regular timeout, tombstone_gc. The use-case motivating this PR is TWCS table, with...
> @denesb - who needs to resolve conflicts here? The author of the PR. I'll tell him to hurry up.
V2: * Rename config item: drop `_compaction` suffix. * Rephrase the config item description, make the connection to `enable_compacting_data_for_streaming_and_repair` clearer. * Rebase.
> why don't we patch sstables_as_snapshot_source() too, to drop expired tombstones with the new "compact_on_read" flag turned one? read repair is another source that can bounce dead tombstones around, right?...
@scylladb/scylla-maint ping.
@scylladb/scylla-maint ping.
> Hi @denesb today it came to my mind what happens given that tombstone being considered expired is no longer deterministic. > > 1st example: repair mode, repairs might have...
> #### 🔴 CI State: FAILURE > > ✅ - [Build](https://jenkins.scylladb.com/job/scylla-master/job/scylla-ci/14427/) ❌ - [Unit Tests Custom](https://jenkins.scylladb.com/job/scylla-master/job/scylla-ci/14427/) **The following new/updated tests ran 100 times for each mode:** 🔹 boost/tablets_test 🔹 topology_custom/test_tablet_repair_scheduler...
> I wonder why we have not seen those failures in the previous CI run. I don't know, but they fail consistently now. There are also conflicts, so this PR...