> The use of threads is also messy in the iotdb-cluster module. Yes.. that is one of the main reasons that I did this.
----------------------------------- Xiangdong Huang School of Software, Tsinghua University 黄向东 清华大学 软件学院 Houliang Qi <neuyi...@163.com> 于2021年8月11日周三 上午9:13写道: > > Decent work! > The use of threads is also messy in the iotdb-cluster module. We can continue > to optimize based on this pr. > > > However, this requires code submitters and reviewers to ensure do not use > `Executors.newXXX()` FUNCTIONS ANY MORE IN THE FUTURE. > > > It would be better if a code detection tool could do this, and I think this > rule can be added to the PR template, so that the PR submitter can see this > rule through self inspection. > > > Thanks, > --------------------------------------- > Houliang Qi > BONC, Ltd > On 08/11/2021 00:37,Xiangdong Huang<saint...@gmail.com> wrote: > Hi, > > We use many thread pools in IoTDB and it is hard to know how many > thread pool we have totally and whether a thread pool is busy. > > Therefore, I unify the creation of ThreadPool, register it to JMX and > allow JMX to check its active size, pending size, etc.. > > When the threadpool is `shutdown()` or `shutdownNow()`, the Pool will > be deregistered from JMX. > > Only The server module has been upgrade to such kind of ThreadPool in > rel/0.12 branch. > > NOTICE, Once this PR is merged, in the server module, DO NOT USE > `Executors.newXXX()` FUNCTIONS ANY MORE IN THE FUTURE. > > For all committers, PLEASE avoid codes like `Executors.newXXX()` when > you review PRs in the future. > > I will do the same thing in the Master branch and replace all thread > pools in the cluster module using this kind of thread pool. > > see PR [1]. > > [1] https://github.com/apache/iotdb/pull/3717 > > Best, > ----------------------------------- > Xiangdong Huang > School of Software, Tsinghua University > > 黄向东 > 清华大学 软件学院