Revert "Revert "rework pushing to views #77309""#79963
Conversation
39731db to
0d77722
Compare
|
I do not know how to fix flaky check for So I wont do it here. |
|
|
| { | ||
| view.runtime_stats->setStatus(QueryViewsLogElement::ViewStatus::QUERY_FINISH); | ||
|
|
||
| LOG_TRACE( |
There was a problem hiding this comment.
Hi @CheSema,
I see that some users find this log useful for various debugging purposes. Do I understand correctly that it was removed prior to the view_duration_ms in system.query_views_log, which shows exactly the same value?
| select_contexts[root_view] = init_context; | ||
| insert_contexts[root_view] = init_context; | ||
| input_headers[root_view] = init_header; | ||
| thread_groups[root_view] = CurrentThread::getGroup(); |
There was a problem hiding this comment.
This will likely not work for background processes, i.e. kafka pushes data to MV, and likely the reason for #86398
There was a problem hiding this comment.
This seems to match our experience - our tables mentioned in that issue is connected to Kafka.
We tried reproducing locally manually inserting data to our views and the events did show up in the system table.
Reverts #79898
Follow up for #77309
I will try to fix discovered errors here.
I hope I had found the reason of the crashes.
Thanks for TSan report here:
https://s3.amazonaws.com/clickhouse-test-reports/json.html?PR=77352&sha=80154a15c7a423a5ce7798a4b741273b5d38ef14&name_0=PR&name_1=Stateless%20tests%20%28tsan%2C%20s3%20storage%2C%201%2F3%29
When that race takes place double free is possible on string reallocation at exception message extension.
Changelog category (leave one):
Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):
This reverts #79898 and restores changes from #77309
Documentation entry for user-facing changes