Thank you to everyone who worked on this - Stamatis, Benchao, Hongyu, Jiajun, and maybe some I’ve missed. It sounded like a very tricky bug to diagnose. (And it wasn’t even a Calcite bug in the end!) A solid, reliable CI system is the foundation of high-quality software.
Gian, The cause seems to have been a subtle bug in Druid’s query cache. If you would like us to log a bug against Druid, let us know. Julian > On Dec 6, 2023, at 8:06 AM, Tanner Clary <tannercl...@google.com.INVALID> > wrote: > > Thanks for fixing this, Stamatis! > > -Tanner > > On Wed, Dec 6, 2023 at 8:04 AM Benchao Li <libenc...@apache.org> wrote: > >> An update on this, it has been solved in CALCITE-6123 by Stamatis, the >> Druid test is supposed to be stable now, thanks Stamatis for fixing >> this and also thanks everyone who helped on this. >> >> Julian Hyde <jhyde.apa...@gmail.com> 于2023年12月2日周六 05:11写道: >>> >>> CI has been failing intermittently for three weeks. In 9 of the last18 >> commits to main, the Druid test has failed. See all the red ‘x’ symbols on >> https://github.com/apache/calcite/commits/main. >>> >>> This has a been logged as >> https://issues.apache.org/jira/browse/CALCITE-6123, and there has been >> plenty of discussion but it hasn’t been resolved. Without functioning CI, >> we are flying blind. There is a high likelihood of more breaking changes >> going in. What should we do? >>> >>> I suggest that we close the main branch until CI is fixed. >>> >>> Julian >>> >> >> >> -- >> >> Best, >> Benchao Li >>