Why MySQL Restarted When Number of Processes Running Now 0 - mysql

Question

Some background. I hosted a WordPress site in VPS and sometimes MYSQL down with error "Error Establishing a Database Connection". I've spent some time to research and believe the problem is due to when MySQL get restarted, it is not able to allocate enough RAM to proceed.
I believe I can improve the situation by increasing physical RAM or swap. But my question for this post is, why MySQL need to restart itself? My site is with pretty low traffic and doesn't seem like the DB is corrupted.
Below is the full log for this issue:
160103 18:39:54 mysqld_safe Number of processes running now: 0
160103 18:39:54 mysqld_safe mysqld restarted
160103 18:39:55 [Note] /usr/libexec/mysqld (mysqld 5.5.44-MariaDB) starting as process 22061 ...
160103 18:39:55 InnoDB: The InnoDB memory heap is disabled
160103 18:39:55 InnoDB: Mutexes and rw_locks use GCC atomic builtins
160103 18:39:55 InnoDB: Compressed tables use zlib 1.2.7
160103 18:39:55 InnoDB: Using Linux native AIO
160103 18:39:55 InnoDB: Initializing buffer pool, size = 128.0M
InnoDB: mmap(137756672 bytes) failed; errno 12
160103 18:39:55 InnoDB: Completed initialization of buffer pool
160103 18:39:55 InnoDB: Fatal error: cannot allocate memory for the buffer pool
160103 18:39:56 [ERROR] Plugin 'InnoDB' init function returned error.
160103 18:39:56 [ERROR] Plugin 'InnoDB' registration as a STORAGE ENGINE failed.
160103 18:39:56 [ERROR] mysqld: Out of memory (Needed 128917504 bytes)
160103 18:39:56 [ERROR] mysqld: Out of memory (Needed 96681984 bytes)
160103 18:39:56 [ERROR] mysqld: Out of memory (Needed 72499200 bytes)
160103 18:39:56 [Note] Plugin 'FEEDBACK' is disabled.
160103 18:39:56 [ERROR] Unknown/unsupported storage engine: InnoDB
160103 18:39:56 [ERROR] Aborting

Answer 1

When Number of Processes Running Now 0 means that MySQL isn't running. So the daemon mysqld_safe "makes you a favor" and start MySQL

Answer 2

You have assigned very low RAM 128MB to innodb_buffer_pool_size (which is default RAM). So you should assign approx. 80% of total RAM to this variable if you are using innodb engine, as mysql uses intially memory from this variable to cache index as well data in innodb engine.
So update at least 1 GB RAM (should be 80% of total ram for innodb) to innodb_buffer_pool_size in your config file and restart mysql service.
Update:
You have 1 GB RAM: You can assign 800M RAM (or >=500M) to innodb_buffer_pool_size.
MySQL will auto restart when number of running process 0: As per errors shared by you...this error is coming at the time of mysql service start and server is not able to initiate mysql service with such a less RAM.
160103 18:39:55 InnoDB: Initializing buffer pool, size = 128.0M
InnoDB: mmap(137756672 bytes) failed; errno 12
160103 18:39:55 InnoDB: Completed initialization of buffer pool
160103 18:39:55 InnoDB: Fatal error: cannot allocate memory for the buffer pool

Answer 3

I had a similar problem with MySQL on CentOS 6.9. It was caused by memory starving due another PHP process and not by MySQL. MySQL was just a victim.
You can look at kernel messages. The location of this messages may be specific to OS. In case of CentOS 6.x they are located in /var/log/messages .
Here are some related messages from /var/log/messages:
Jul 25 20:34:46 myserver kernel: Out of memory: Kill process 21467 (mysqld) score 30 or sacrifice child
Jul 25 20:34:46 myserver kernel: Killed process 21467, UID 497, (mysqld) total-vm:757004kB, anon-rss:17728kB, file-rss:320kB
You can run the following command to see if kernel run out of memory:
cat /var/log/messages | grep out_of_memory