Il 10/09/2023 13:56, Sergei Golubchik ha scritto:
that's recent enough, should have everything fixed already. May be a new bug. Could you please report it at jira.mariadb.com?
Few questions to cover there (someone will ask them, so you might as well know what's coming and may be even answer in advance).
So, process size is growing? Like
ps -o pid,vsz,comm `pgrep mysqld`
I have rebooted my vps for a linux upgrade just 5 hours ago and I haven't even used roundcube yet. the command you posted isn't right for me, "pgrep mysqld" returns nothing. But this work: ps -o pid,vsz,comm `pgrep mariadb` PID VSZ COMMAND 832 1073640 mariadbd Also, running the following command to check swap: for file in /proc/*/status ; do awk '/VmSwap|Name/{printf $2 " " $3}END{ print ""}' $file; done | sort -k 2 -n -r | less mariadbd is currently the first entry: mariadbd 90624 kB
is growing? What does
show global status like 'memory%' select id,memory_used from information_schema.processlist;
show? Do you have long-running connections? If you disconnect them all, how the above will change?
+---------------------+----------+ | Variable_name | Value | +---------------------+----------+ | Memory_used | 37291824 | | Memory_used_initial | 35240160 | +---------------------+----------+ 2 rows in set (0.018 sec) +----+-------------+ | id | memory_used | +----+-------------+ | 31 | 78096 | +----+-------------+ 1 row in set (0.013 sec) Maybe I will wait a couple days and check the differences, too early to say I think. Thanks