I can't reproduce this on my MediaWiki server, but it's very reproducible on my Moodle server. If I do: systemctl reload httpd.service or if anything does that (like a package upgrade that calls the httpd filetrigger), after that point, every connection made to https on the server will cause the child handling that connection to segfault (visible in the error log) and you get an error message in the browser. Connecting to the server with just http still works fine. Upgrading to a local build of Apache 2.4.9 doesn't help, so I think it's an issue with the systemd service file. Perhaps this is related: https://bugzilla.redhat.com/show_bug.cgi?id=952634 Should we be using the KillMode=mixed that Lennart mentioned? Reproducible: Steps to Reproduce:
CC: (none) => mageia, oe
This is really strange, even if you bypass systemd and just manually send the main httpd process a USR1 or HUP signal (or use httpd -k restart or httpd -k graceful), it causes the same issue. So this would seem to be an upstream issue and not a systemd one, but how could upstream not have fixed this?
I also just noticed that loading /index.html through https works fine, it's just hitting Moodle itself (aka PHP) that segfaults.
OK, this isn't actually https-specific after all. Status pages still work, just PHP ones don't. So mod_php appears to have a problem if the main httpd process has re-read its configuration.
Summary: apache mod_ssl segfaults after reloading the service => apache-mod_php segfaults after reloading the service
s/Status/Static/ (sorry)
So this is a duplicate of Bug 12995 after all. *** This bug has been marked as a duplicate of bug 12995 ***
Status: NEW => RESOLVEDResolution: (none) => DUPLICATE