Question Innodb crashes

[LucaB.]

Hi,
I've some alerts in mariadb.log (Centos7 12.5.30 #62 2CPU 2GB ram, nginx + apache with php-fpm serves ~70 websites):
170218 16:52:29 InnoDB: Database was not shut down normally!
170320 8:42:48 InnoDB: Database was not shut down normally!
170321 2:07:52 InnoDB: Database was not shut down normally!
170327 7:27:09 InnoDB: Database was not shut down normally!
170327 16:16:54 InnoDB: Database was not shut down normally!

Last log reports:
170327 16:15:49 mysqld_safe Number of processes running now: 0
170327 16:15:56 mysqld_safe mysqld restarted
170327 16:16:44 [Note] /usr/libexec/mysqld (mysqld 5.5.52-MariaDB) starting as process 30986 ...
170327 16:16:49 InnoDB: The InnoDB memory heap is disabled
170327 16:16:49 InnoDB: Mutexes and rw_locks use GCC atomic builtins
170327 16:16:49 InnoDB: Compressed tables use zlib 1.2.7
170327 16:16:49 InnoDB: Using Linux native AIO
170327 16:16:50 InnoDB: Initializing buffer pool, size = 128.0M
170327 16:16:50 InnoDB: Completed initialization of buffer pool
170327 16:16:53 InnoDB: highest supported file format is Barracuda.
InnoDB: The log sequence number in ibdata files does not match
InnoDB: the log sequence number in the ib_logfiles!
170327 16:16:54 InnoDB: Database was not shut down normally!
InnoDB: Starting crash recovery.
InnoDB: Reading tablespace information from the .ibd files...
InnoDB: Restoring possible half-written data pages from the doublewrite
InnoDB: buffer...
170327 16:18:18 InnoDB: Waiting for the background threads to start
170327 16:18:19 Percona XtraDB (Experts in Database Performance Management) 5.5.49-MariaDB-38.0 started; log sequence number 9422764659
170327 16:18:19 [Note] Plugin 'FEEDBACK' is disabled.
170327 16:18:20 [Note] Server socket created on IP: '::'.
170327 16:18:20 [Warning] 'proxies_priv' entry '@ [email protected]' ignored in --skip-name-resolve mode.
170327 16:18:22 [Note] Event Scheduler: Loaded 0 events
170327 16:18:22 [Note] /usr/libexec/mysqld: ready for connections.

What can I do to avoid these mistakes?

Thanks.


Luca

 

[UFHH01]

Hi LucaB.,

InnoDB: The log sequence number in ibdata files does not match
InnoDB: the log sequence number in the ib_logfiles!

This indicates, that you previously didn't shutdown your MySQL - server correctly. Often enough, this can as well happen, if your ( v-server? ) lacks enough memory.

2CPU 2GB ram, nginx + apache with php-fpm serves ~70 websites

As you can see, you don't really have much memory installed. Consider to invest something here.

InnoDB: Database was not shut down normally!

Because of your crash(es), your InnoDB - tables have to be recovered. As InnoDB is designed to do this AUTOMATICALLY you should't worry here, but you should be aware, that the recovery may take some time. Often enough it's only seconds/minutes to recover the tables, but in case that you had a few hundreds pages modified data in your memory, the process can as well take some hours.
The recovery itself will only start, if you setup the "recovery mode" in your "my.cnf". Pls. see: => XtraDB/InnoDB Recovery Modes , to inform yourself about possible settings!

Steps to go:

insert "innodb_force_recovery = 0" to your "my.cnf" as a start and inspect your error - log. If "0" isn't already enough, consider to set the next level => "1" and again: pls. inspect your error - log.​