Liberating my Server from (MariaDB) Database Logs

My server running Docker seems forever to be low on space. So I finally got annoyed at it enough to figure out what was going on and try and fix it.

First, I needed to figure out what was taking up soooo much space. ncdu — which was even already installated! — was the tool of choice. And it promptly found 63.8GB (on a 256GB drive) in docker/volumes/homeassistant-db/logs/mysql; all binary logs.

I have logs going back over a year, but I never look at them. So how best to clean these up?

First, I looked into automatic log rotation, which should keep some of the newer logs (allowing me to look for recent issues, if needed) but automatically remove older ones. Looking at the config file, it seems to be set up right. The config file is at docker/volumes/homeassistant-db/custom.cnf and the lines of interest:

129
130

expire_logs_days        = 10
max_binlog_size         = 100M

This is saying that the log files have a (soft) size limit of 100 MB (which seems to be being respected) and that logs older than 10 days should automatically be deleted (which isn’t being respected).

Ok, let’s see if I can manually rotate the logs. To do this, I need a shell in the MariaDB Docker container; I have Portainer set up to allow me to do this. Once at the shell, I run mariadb to get a prompt to directly edit the database. From there, I’ll drop everything older than ~1 week:

MariaDB [(none)]> PURGE BINARY LOGS BEFORE '2025-11-22';
Query OK, 0 rows affected, 1 warning (0.005 sec)

This looks, at first glance, like it worked. However, the files are still on disk, and there is a notation about one warning. Okay, what’s the warning?

MariaDB [(none)]> SHOW WARNINGS;
+-------+------+----------------------------------------------------------------------------+
| Level | Code | Message                                                                    |
+-------+------+----------------------------------------------------------------------------+
| Note  | 1375 | 'mariadb-bin.000570' is not purged because it is the current active binlog |
+-------+------+----------------------------------------------------------------------------+
1 row in set (0.001 sec)

Ah, no it’s not the “active binlog” (or it shouldn’t be); that’s our first log file still on disk, and is over a year old! But if the database server thinks that the case, that would explian why it’s keep all the logs since.

But I can dump all the logs, since I’ve never referred to them anyway.

MariaDB [(none)]> RESET MASTER TO 1260;
Query OK, 0 rows affected (0.058 sec)

The TO 1260 is optional; it sets the number that the new log files should start with. I set it here just a couple higher than the previous most recent log file (1257).

Success! The log files are gone!

Next up: add a sensor to Home Assistant to monitor the size of this folder…

References:

• expire_log_days in the MariaDB configuration documentation
• SHOW BINARY LOGS on MariaDB documentation
• PURGE BINARY LOGS on MariaDB documentation
• SHOW WARNINGS on MariaDB documentation
• RESET MASTER on MariaDB documentation
• Error 1375 in the MariaDB Error Codes Reference