This is a known issue in DM 2.0.0 version and will be fixed in DM 2.0.1 version. It is likely to be triggered when a replication task has a lot of tables to process. If you use TiUP to deploy DM, you can upgrade DM to the nightly version to fix this issue. Or you can download the 2.0.0-hotfix version from the release page of DM on GitHub and manually replace the executable files.