The apt preferences files you created are being ignored because of their name (apt thinks .7 is the extension, so they would need to end with .pref). Here's the manpage
$ sudo apt update
[snip]
N: Ignoring file 'mysql-client-core-5.7' in directory '/etc/apt/preferences.d/' as it has an invalid filename extension
N: Ignoring file 'mysql-client-5.7' in directory '/etc/apt/preferences.d/' as it has an invalid filename extension
N: Ignoring file 'mysql-server-core-5.7' in directory '/etc/apt/preferences.d/' as it has an invalid filename extension
N: Ignoring file 'mysql-server-5.7' in directory '/etc/apt/preferences.d/' as it has an invalid filename extension
The mysql packages were updated again on Saturday (/var/log/unattended-upgrades/unattended-upgrades-dpkg.log), but this time the mysql service did not restart, so nothing broke
$ sudo systemctl status mysql.service
[snip]
Active: active (running) since Fri 2019-07-26 13:24:16 UTC; 2 days ago
[snip]
I'm trying to make sense of the logs from Friday when it broke after the update (/var/log/mysql/error.log.3.gz), however there are no errors in them, just the daemon shutting down right after starting
The apt preferences files you created are being ignored because of their name (apt thinks .7 is the extension, so they would need to end with .pref). Here's the [manpage](https://manpages.debian.org/jessie/apt/apt_preferences.5.en.html)
```
$ sudo apt update
[snip]
N: Ignoring file 'mysql-client-core-5.7' in directory '/etc/apt/preferences.d/' as it has an invalid filename extension
N: Ignoring file 'mysql-client-5.7' in directory '/etc/apt/preferences.d/' as it has an invalid filename extension
N: Ignoring file 'mysql-server-core-5.7' in directory '/etc/apt/preferences.d/' as it has an invalid filename extension
N: Ignoring file 'mysql-server-5.7' in directory '/etc/apt/preferences.d/' as it has an invalid filename extension
```
The mysql packages were updated again on Saturday (`/var/log/unattended-upgrades/unattended-upgrades-dpkg.log`), but this time the mysql service did not restart, so nothing broke
```
$ sudo systemctl status mysql.service
[snip]
Active: active (running) since Fri 2019-07-26 13:24:16 UTC; 2 days ago
[snip]
```
I'm trying to make sense of the logs from Friday when it broke after the update (`/var/log/mysql/error.log.3.gz`), however there are no errors in them, just the daemon shutting down right after starting
On andromeda there is a /etc/systemd/system/mysql.service file that was created by Chef, on a fresh VM the service is provided by the package and is located at /lib/systemd/system/mysql.service. So it looks like an issue caused by past versions of the mysql cookbook
Installing the mysql-server-5.7 package again and disabling the /etc/systemd/system/mysql.service service should do the trick. I'm going to announce on IRC/XMPP that the wiki might go down for a bit
On andromeda there is a `/etc/systemd/system/mysql.service` file that was created by Chef, on a fresh VM the service is provided by the package and is located at `/lib/systemd/system/mysql.service`. So it looks like an issue caused by past versions of the mysql cookbook
Installing the `mysql-server-5.7` package again and disabling the `/etc/systemd/system/mysql.service` service should do the trick. I'm going to announce on IRC/XMPP that the wiki might go down for a bit
I have managed to update mysql-server-5.7, after stopping /etc/systemd/system/mysql.service, moving it, running systemctl daemon-reload.
I had to temporarily add the UNIX socket /run/mysql/mysqld.pid to the /var/www/mediawiki-1.32.0/LocalSettings.php, because the config in /etc/mysql was also generated by an old version of the mysql cookbook, and that set the UNIX socket to the non-standard /run/mysql/mysqld.pid. The one set by the mysql package is /var/run/mysqld/mysqld.sock, that is a default known by Mediawiki and other apps. So now the final step is to restore the initial config from the packages
I have managed to update mysql-server-5.7, after stopping `/etc/systemd/system/mysql.service`, moving it, running `systemctl daemon-reload`.
I had to temporarily add the UNIX socket `/run/mysql/mysqld.pid` to the `/var/www/mediawiki-1.32.0/LocalSettings.php`, because the config in `/etc/mysql` was also generated by an old version of the mysql cookbook, and that set the UNIX socket to the non-standard `/run/mysql/mysqld.pid`. The one set by the mysql package is `/var/run/mysqld/mysqld.sock`, that is a default known by Mediawiki and other apps. So now the final step is to restore the initial config from the packages
I have restored the content of /etc/mysql, restarted the service, and restored the mysql config in /var/www/mediawiki-1.32.0/LocalSettings.php to what it was before (removed the :/run/mysql/mysqld.pid suffix from the $wgDBserver definition, setting it back to "localhost"). Also removed the valid (libmysqlclient20 & libmysqlclient-dev) and invalid (all the mysql-*-5.7) apt preferences
This is all done and working now.
I have restored the content of `/etc/mysql`, restarted the service, and restored the mysql config in `/var/www/mediawiki-1.32.0/LocalSettings.php` to what it was before (removed the `:/run/mysql/mysqld.pid` suffix from the `$wgDBserver` definition, setting it back to "localhost"). Also removed the valid (libmysqlclient20 & libmysqlclient-dev) and invalid (all the mysql-*-5.7) apt preferences
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
See #82 for more details. They are currently pinned to a specific version.
The apt preferences files you created are being ignored because of their name (apt thinks .7 is the extension, so they would need to end with .pref). Here's the manpage
The mysql packages were updated again on Saturday (
/var/log/unattended-upgrades/unattended-upgrades-dpkg.log), but this time the mysql service did not restart, so nothing brokeI'm trying to make sense of the logs from Friday when it broke after the update (
/var/log/mysql/error.log.3.gz), however there are no errors in them, just the daemon shutting down right after startingOn andromeda there is a
/etc/systemd/system/mysql.servicefile that was created by Chef, on a fresh VM the service is provided by the package and is located at/lib/systemd/system/mysql.service. So it looks like an issue caused by past versions of the mysql cookbookInstalling the
mysql-server-5.7package again and disabling the/etc/systemd/system/mysql.serviceservice should do the trick. I'm going to announce on IRC/XMPP that the wiki might go down for a bitI have managed to update mysql-server-5.7, after stopping
/etc/systemd/system/mysql.service, moving it, runningsystemctl daemon-reload.I had to temporarily add the UNIX socket
/run/mysql/mysqld.pidto the/var/www/mediawiki-1.32.0/LocalSettings.php, because the config in/etc/mysqlwas also generated by an old version of the mysql cookbook, and that set the UNIX socket to the non-standard/run/mysql/mysqld.pid. The one set by the mysql package is/var/run/mysqld/mysqld.sock, that is a default known by Mediawiki and other apps. So now the final step is to restore the initial config from the packagesThis is all done and working now.
I have restored the content of
/etc/mysql, restarted the service, and restored the mysql config in/var/www/mediawiki-1.32.0/LocalSettings.phpto what it was before (removed the:/run/mysql/mysqld.pidsuffix from the$wgDBserverdefinition, setting it back to "localhost"). Also removed the valid (libmysqlclient20 & libmysqlclient-dev) and invalid (all the mysql-*-5.7) apt preferences