There was a lot of discussion about the code in the PR with many good questions and a lot of back and forth. There was a lot of discussion and questions about the concept of how a set of files is selected for user compactions and what this means for subsequent system compactions. Given the amount of discussion around this I will attempt to document this well in the user manual. There was a request to update the example shell commands to use this new feature and I plan to do that and will link it on the PR.
The PR is large and I intentionally held back on making some changes. I was asking for thoughts about opening follow on issues and Christopher suggested creating a github project to track the issues. I thought that was a great idea and created : https://github.com/apache/accumulo/projects/13 I have already created a few issues under the project and plan to create a few more. One thing I requested in the slack chat is for someone to review the file CompactableImpl.java in the PR as this a very important bit of code that sits between compaction services and a tablet. Ed suggested making some file stats like creation time available for compaction selectors. This needs a follow on issue. I mentioned that per compaction service metrics are needed and Ed had some good ideas around this area. I will open an issue and maybe Ed could add to that. On Wed, May 13, 2020 at 9:41 AM RawDNACoders PurePeace <[email protected]> wrote: > > I am OOO sick today. I would love to get a playback or notes following the > discussion. Can someone please take notes and share? > Thank you! > > On Mon, May 11, 2020, 10:35 AM Keith Turner <[email protected]> wrote: > > > Christopher suggested a chat on slack to discuss the following. For > > anyone who is interested in joining, would Wed 5/13 at 1300 EST work? > > If not, we can find another time once we know who is interested. > > > > https://github.com/apache/accumulo/pull/1605 > > > > Keith > >
