Noticed in bug #175... There are a LOT of places where slon_retry() is called, which signals a start of the slon daemon. local_listener.c: 8 cleanup_thread.c: 4 remote_worker.c: 45 sync_thread.c: 5 misc.c: 3 runtime_config.c: 15 slon.c: 15 This seems rather excessive - if a problem occurs in any of these places, this effectively terminates all activity that's going on. If a 2-day-long subscription is taking place, then any little problem that occurs in a peripheral area will take that down and restart it. Some assessment should take place to determine the degree to which this is excessive. There are legitimately problems that are fatal that warrant terminating/restarting the slon daemon. But probably not *quite* this many...