You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
after several times doing this: systemctl reload httpd
php no longer has mysqli support, i.e. it no longer shows in phpinfo() and functions like mysqli_report() are no longer defined.
Note the difference between "reload" and "restart".
Several reloads break it. One restart gets it working again.
Steps To Reproduce
Steps to reproduce the behavior:
seq 1 40 | while read x; do systemctl reload httpd; done
The number of reloads it takes to break php+mysql varies. Perhaps it takes (on average) less reloads with a larger httpd.conf, so if you have trouble reproducing, perhaps try a few hundred reloads.
Please let me know if you have trouble reproducing and I'll work on creating a simpler test case where I can publish my full configuration.nix
Additional context
I have 24 virtualhosts configured, all with acme. I'm seeing an httpd reload on average about once per hour, so php breaks about once per day. I can see httpd reload with: systemctl status -n 200 httpd
I suspect that the acme timers are causing a lot of unnecessary httpd reloads, but that's a separate issue.
I just remembered that php81 is now available, so I tested with that and was unable to reproduce this bug with php81 instead of php80 as shown in my configuration.nix sample above.
So, apologies for pinging apache maintainers. I'll attempt to ping a couple php folks:
Describe the bug
after several times doing this:
systemctl reload httpd
php no longer has mysqli support, i.e. it no longer shows in phpinfo() and functions like mysqli_report() are no longer defined.
Note the difference between "reload" and "restart".
Several reloads break it. One restart gets it working again.
Steps To Reproduce
Steps to reproduce the behavior:
seq 1 40 | while read x; do systemctl reload httpd; done
echo "<?php mysqli_report(0); print 'OK';" > /your/web/root/test.php
curl http://localhost/test.php
Expected behavior
curl should output "OK".
The number of reloads it takes to break php+mysql varies. Perhaps it takes (on average) less reloads with a larger httpd.conf, so if you have trouble reproducing, perhaps try a few hundred reloads.
Please let me know if you have trouble reproducing and I'll work on creating a simpler test case where I can publish my full configuration.nix
Additional context
I have 24 virtualhosts configured, all with acme. I'm seeing an httpd reload on average about once per hour, so php breaks about once per day. I can see httpd reload with:
systemctl status -n 200 httpd
I suspect that the acme timers are causing a lot of unnecessary httpd reloads, but that's a separate issue.
Notify maintainers
@winterqt @aanderse
Metadata
Please run
nix-shell -p nix-info --run "nix-info -m"
and paste the result.This is running on a $5/month Linode.
Here's how I configure apache/php:
my 24 virtualhosts all have:
The text was updated successfully, but these errors were encountered: