On Tue, Apr 21, 2020 at 10:07:27AM +0200, Domenico Briganti wrote: > Thanks William, yes, the reload of haproxy is a feasible way, I hadn't > noticed.I have just one doubt, since I update the crl every day and I > have mqtt connections that can stay connected for days, at the end I > can have many haproxy process running, one a day, until all old > connection (of that day) terminates. I think that with ps and netstats > I can see how many they are and how many old connections each process > manages.However I can afford a complete restart of haproxy once every > two/three weeks. > Regards,Domenico
If you configure the master CLI (haproxy -S binary argument), you will be able to access to the CLI of the previous process and monitor the remaining connections. The previous process won't leave until the connections aren't closed. You can force a process to leave even if there are still some connections with the directive "hard-stop-after". https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-hard-stop-after You can also limit the number of workers with the directive "mworker-max-reloads". https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-mworker-max-reloads Regards, -- William Lallemand