Items tagged with: mysql
anyone any idea what this messages in journalctl mean? I got lots of these before my server crashed and have no Idea what this is.
Mai 28 03:10:26 friendica mariadbd: 2021-05-28 3:10:26 0 [Warning] InnoDB: A long semaphore wait: Mai 28 03:10:26 friendica mariadbd: --Thread 139676422764096 has waited at row0upd.cc line 179 for 465.00 seconds the semaphore: Mai 28 03:10:26 friendica mariadbd: S-lock on RW-latch at 0x55a2a8136178 created in file dict0dict.cc line 1041 Mai 28 03:10:26 friendica mariadbd: a writer (thread id 139674262697536) has reserved it in mode wait exclusive Mai 28 03:10:26 friendica mariadbd: number of readers 1, waiters flag 1, lock_word: ffffffff Mai 28 03:10:26 friendica mariadbd: Last time write locked in file dict0stats.cc line 2487
I guess something was wrong with either MariaDB or the VM itself that cause long operation times on the database which led to lots of long running locks. But what do I do with this?
#Friendica #MariaDB #Mysql #Linux #Semaphore
mysqltuner.pl wants me to set "innodb_buffer_pool_size (>= 20.7G)", which is probably because of the #database size. I assigned 20 GB #RAM to the VM and have 20 GB #SWAP too. So setting it that high is not possible, as this would bring the total max #memory usage way beyond 40 GB.
1. Why is the total max usage higher than this setting? What is the multiplicand parameter?
2. Any idea how I can set this to use 16 GB for #MySQL all the time, allow double in peak and (swapping then) and not more?
I think I will change the setup at some point which will free more RAM for #mariadb, but now I need to cope with it like it is. I see loading time it really bad with low settings for "innodb_buffer_pool_size", like 0.4 seconds over all, more RAM reduces it by 70%... but then I get omm (out of memory) problem.