Hi all, We're running some tornado based servers from supervisor, and we are having a strange behavior when trying to restart the tornado instances:
After the process is killed (either with SIGTERM or SIGKILL), sometimes it stays in the STOPPED status for a long time (sometimes up to 25 minutes). In the meantime, the tornado process is already gone (it doesn't show in ps anymore), and supervisor doesn't report anything to the log file between the stopped messaage and the one about starting the process again (after this random time). I've tried disabling logging of the process' stderr and stdout, and it seemed to work a bit faster, but we still have this issue. How can I debug what supervisor is doing between the STOPPED and the START status changes? Could it be that something remains open and supervisor is waiting for some timeout? In that case, is that timeout configurable? Thanks -- Gabriel E. Patiño http://about.me/gepatino
_______________________________________________ Supervisor-users mailing list [email protected] https://lists.supervisord.org/mailman/listinfo/supervisor-users
