On 2020-06-23 22:48, David G. Johnston wrote:
In the following log file the presence of "exit code 1" after performing a "pg_ctl stop -m smart" shutdown is bugging me. I take it most people would just ignore it as noise but a clean install from source, startup, and shutdown would ideally not result in a non-zero exit code being sent to the log. But I've decided to stop trying to track it down on my own and at least mention it here. It seems like f669c09989 probably introduced the behavior, and I can see how "restarting" is generally a good thing, but we are not going to restart the launcher during a clean shutdown and the exit code isn't being done conditionally.
Yeah, this is just a consequence of how background workers work. One would need to adjust the handling of signaling, restart flags, exit codes, etc. to be able to do this more elegantly. It's probably not that hard, but it needs some leg work to go through all the cases and issues.
-- Peter Eisentraut http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
