Bad configuration MariaDB / MySQL

I currently have a LAMP on a dedicated server but with serious problems with MariaDB (MySQL).

I have installed a website with 100k daily pageviews and due to the need for the queries it is necessary for the database server to have a cache or the site would be very slow.

The problem is that I have a bad configuration in the parameters of the file:

/etc/my.cnf.d/server.cnf

The server goes up to 100% CPU and 100% memory, does not save cache and crash.

Hardware
Server type: Dedicated
CPU: Xeon E5-2650 v4 (x2)
RAM: 128GB
SSD: 2T

Software 
Os: CentOS 7
Server: MariaDB v10.11.4
Tables: 66
Rows: 346,215,143
Engine: innoDB
Size data base: 74,837 MB
Collation database: utf8mb4_general_ci
Callation tables: utf8mb4_general_ci

This is the file:

[server]

[mysqld]
datadir                         = /var/lib/mysql
socket                          = /var/lib/mysql/mysql.sock
log-error                       = /var/log/mariadb/mariadb.log
pid-file                        = /var/lib/mysql/hippo.pid
port                            = 5000
#skip-networking
symbolic-links                  = 0
bind-address                    = 0.0.0.0
default_authentication_plugin   = mysql_native_password
skip-name-resolve               = 1 
max_connections                 = 15000
max_user_connections            = 15000
query_cache_type                = 1
query_cache_size                = 64G
query_cache_limit               = 2G
innodb_file_per_table           = ON
innodb_buffer_pool_size         = 46G
innodb_flush_method             = O_DIRECT
innodb_read_io_threads          = 16
innodb_write_io_threads         = 16
innodb_io_capacity              = 5000
open_files_limit                = 32768
innodb_open_files               = 32768
innodb_log_buffer_size          = 32M
innodb_log_file_size            = 19G
innodb_data_file_path           = ibdata1:10M:autoextend
innodb_flush_log_at_timeout     = 3
key_buffer_size                 = 16G
sort_buffer_size                = 16G
read_buffer_size                = 16G
read_rnd_buffer_size            = 16G
join_buffer_size                = 128M 
thread_cache_size               = 64
tmp_table_size                  = 32G
max_heap_table_size             = 32G
wait_timeout                    = 300
net_read_timeout                = 300
net_write_timeout               = 600
table_open_cache                = 32768

[mysqld_safe]
open_files_limit                = 32768

[galera]

[embedded]

[mariadb]

[mariadb-10.3]

Thank you :)

A few pages per second should not be a problem. (1K/second could be a problem.)

Many settings are threatening to overflow your large memory and should be shrunken:

max_connections                 = 15000  --> 200
max_user_connections            = 15000  --> 200
query_cache_size                = 64G  --> 100M
query_cache_limit               = 2G   --> 50M
innodb_buffer_pool_size         = 46G  --> 70G (yes, increase)
key_buffer_size                 = 16G  --> 50M (if not using MyISAM)
sort_buffer_size                = 16G  --> 1G
read_buffer_size                = 16G  --> 1G
read_rnd_buffer_size            = 16G  --> 1G
tmp_table_size                  = 32G  --> 1G
max_heap_table_size             = 32G  --> 1G

The "Query cache" is not very efficient when it is big.

Each connection consumes lots of RAM, so 15000*"lots" is possibly causing your memory problem. Since you deliver only a few pages per second, you don't need more than a few connections.

The buffer_pool is the may caching mechanism; most of RAM should be given to it. It you hit 100% RAM, decrease it further.

Turn on the SlowLog to see which queries are the slowest; then let's look at the indexes involved.

If you are running a web server (Apache?) on the same server, maybe it is over-configured, too.

This really is a comment, but its a bit too long to fit in the wee box.

I have installed a website with 100k daily pageviews

This is a meaningless metric.

does not save cache and crash

What do you mean it crashes? Core dumps? OOM Killer? Something else?

You certainly have capacity issues - how you fix those takes a lot more time, information and effort than a quick post here.

If your memory usage is getting to 100% then your configuration is wrong - but not necessarily your database config. While the max connections for mysql is much too high, you need to address this on your webserver.

Whether the other settings in your database are appropriate or wrong is IMPOSSIBLE to tell without a detailled understanding of the workload.