Nikita Taranov
Nikita Taranov
I tested on TCP-H 10, Q6. there is some noticeable slowdown though not really big (slightly less 10%) ``` Queries executed: 787. localhost:9000, queries 787, QPS: 10.491, RPS: 629283976.156, MiB/s:...
#40142 covers the case we observed on TPCH Q6. I found one more case that will be addressed in subsequent pr.
just to double-check. if we join large table and small table, the large table's set will overflow and won't be used. the small table's set will reach finished state after...
tests with projections will be fixed in #39671
02405_pmj_issue_40335 - #40610
some of the possible bad cases from the chat with Vladimir https://pastila.nl/?009cde16/669061285f5478a836a4837d2b41ca00 added just to not be lost from the context
`distinct_in_order` perf test seems to be unstable
I also would like to mention in the comments that we need to handle discussed case with aliases, just to not forgot about it.
> AST fuzzer failures are related, but I have no ideas why fixing an AST optimization leads to "pipeline stuck"... > > @nickitat, could you please help? I will take...
similar report: https://s3.amazonaws.com/clickhouse-test-reports/41247/a963fbd653e3d9924df73aa334b0159e13fea5ed/stateless_tests__tsan__s3_storage__[1/3]/stderr.log looks like `onRename` and `getFullPath` called concurrently and we just don't synchronise access to `part_dir` and `root_path` with a lock.