Botond Dénes

Seen again in https://jenkins.scylladb.com//job/scylla-master/job/scylla-ci/22549/testReport/junit/internode_compression_test/TestInternodeCompression/release___dtest_with_consistent_topology_changes___test_internode_compression_between_datacenters

Seen again in https://jenkins.scylladb.com/job/scylla-master/job/scylla-ci/22407/testReport/cluster/test_cluster_features/debug___test_py___debug___test_joining_old_node_fails_debug_2/

Seen again in https://jenkins.scylladb.com/job/scylla-master/job/scylla-ci/22404/testReport/junit/cluster/test_tablets/debug___test_py___debug___test_two_tablets_concurrent_repair_and_migration_repair_writer_level_debug_3/.

This is the condition variable which times out: ``` 561 return utils::get_local_injector().inject("repair_writer_impl_create_writer_wait", utils::wait_for_message(200s)); ``` Looks like 200s is not enough anymore.

Promtion is broken currently, I added merge_candidate label and I will come back and queue it once it is restored.

> > Promtion is broken currently, I added merge_candidate label and I will come back and queue it once it is restored. > > What is broken? Where was this...

I'm on gcc-8.1 and seastar compiles fine for me, though I only compile `libseastar.a`.

With our current way of estimating partition count of output sstables for compaction, any workload which is delete or overwrite heavy, and thus can result in big partition-count reduction when...

Need similar fix for https://github.com/scylladb/scylladb/blob/a213e41250dff8211ac1b539b0f4ed1910202801/streaming/stream_session.cc#L90

> > Need similar fix for > > Yes, I saw it. But I could not verify it was related to the OP repair path...? Right. Streaming & tablets don't...