commit | 2b9de8d4f80dc13613b2545076d3fda8b0d8bf3c | [log] [tgz] |
---|---|---|
author | WEI Xikai <ShiKaiWi@users.noreply.github.com> | Tue May 30 14:37:18 2023 +0800 |
committer | GitHub <noreply@github.com> | Tue May 30 14:37:18 2023 +0800 |
tree | b33d9c38122c5fe90a370c226dc3d2ffa69513d8 | |
parent | 9058868a451198e7a093c071cef85ed951f67e2d [diff] |
fix: write cancel when flush pending write queue (#940) ## Related Issues Closes # ## Detailed Changes Currently, the writer for the first request in the pending write queue is responsible for flushing all the writes in the queue, but it can't be ensured to always trigger flush because it may be cancelled when await for the serial exec lock. And this change set replaces the `try_join_all` in the rpc services to avoid such cancellation. ## Test Plan Test it manually.
CeresDB is a high-performance, distributed, cloud native time-series database.
Any contribution is welcome!
Read our Contributing Guide and make your first contribution!
Some design of CeresDB references influxdb_iox, and some specific module implementations reference tikv and other excellent open source projects, thanks to InfluxDB, TiKV, and any other referenced great open source projects.
CeresDB is under Apache License 2.0.