To take a checkpoint that's used for coordinated restore, all file handles and network sockets most be closed. This will require the logging system's appenders to be stopped/paused as they may be writing to files or sending log events over the network. Ideally, this would be done by the Logback or Log4j2 participating directly in the checkpointing process but that may not work out. We need to evaluate our options and figure out what to do.