Pavel Borisov

at 9c34d00d2615 Bucket size after inserts ``` data/ 5.1 GB / 1209 files orioledb_data/ 4.0 GB / 100233 files wal/ 5.2 GB / 335 files ``` After 16 start-stops: ```...

Rechecked. orioledb_undo bloating disappeared. Also, it's very nice that cluster start and stop times decreased significantly. First cluster stop ~50s, then start ~10s, then all start-stops (with no data inserted...

On s3 bucket dir structure: - /data//orioledb_undo is empty (maybe doesn't need to be created) - Orioledb part of checkpoint is split into /data//orioledb_data (1) and /orioledb_data/ (2) (maybe move...

I see several tests fail under valgrind. It seems like Postgres is stuck at stopping. I couldn't see this fail on the other branches, though there were some valgrind timeout...

Tests of this PR passed. I think it's ready to be merged.

Maybe related to https://github.com/orioledb/orioledb/issues/351 but in this ticket I didn't intentionally corrupted anything.

With 50 s3 workers insert performance don't differ, though load per each s3 worker is around 2 times less

at c34d00d2615d8ee14 insert performance after start-stops increased significantly: First insert: ``` progress: 10.0 s, 33246.5 tps, lat 1.499 ms stddev 1.967, 0 failed progress: 20.0 s, 30935.1 tps, lat 1.615...

at a1d732000f74b1011 I checked insert performance again and the drop in insert performance is slight: - After many hours of inserts - After a couple of start-stops Backtrace of insert...

- With 10000 partitions and ~25 Gb cluster it takes ~4.5 minutes to start/stop cluster for both s3 and non-s3 mode - With 100 partitions and ~25 Gb it takes...