I fell in love with Virtualmin (coming from WHM) quite recently and just moved all websites with low traffic to our new virtualmin powered server.
From the performance point of view, the server is amazing. All websites are really snappy.
However, the RAM usage is crazy (in my opinion). It usually averages around 80%, at the same time the highest CPU usage I saw so far is 24%.
Here are the server specifications:
- Operating system: CentOS Linux 7.4.1708
- Webmin version: 1.881
- Usermin version: 1.741
- Virtualmin version: 6.03
- Theme version: Authentic Theme 19.09.2
- Kernel and CPU: Linux 3.10.0-693.21.1.el7.x86_64 on x86_64
- Processor information: Virtual CPU 82d9ed4018dd, 4 cores
- System uptime: 4 days, 21 hours, 49 minutes
- Running processes: 337
- CPU load averages: 0.00 (1 min) 0.04 (5 mins) 0.05 (15 mins)
- Real memory: 5.60 GB used / 7.46 GB total
- Virtual memory: 27.59 MB used / 3.91 GB total
- Local disk space: 23.64 GB used / 74.80 GB free / 98.43 GB total
- I installed NGINX
- I installed PHP 7.2
- I update MariaDB to 10.2
- I installed REDIS and configured it to save PHP sessions
- I configured 4GB swap space with a swappiness of 30
When I click on the memory bar in Webmin, it shows mostly NGINX processes (4 to 6 per user), however, htop via SSH outputs this:
So based on htop it looks like MySQL is the culprit. I already tried to reconfigure MySQL (MariaDB). MySQL tuner outputs this:
Maximum possible memory usage: 2.9G (38.00% of installed RAM)
Which sounds reasonable to me.
For completion this is my MySQL configuration:
innodb_file_per_table = 1
thread_concurrency = 8
query_cache_size = 0
query_cache_type = 0
thread_cache_size = 8
myisam_sort_buffer_size = 16M
read_rnd_buffer_size = 4M
read_buffer_size = 2M
sort_buffer_size = 2M
table_open_cache = 512
max_allowed_packet = 1M
key_buffer_size = 32M
key-buffer-size = 4M
max-allowed-packet = 2M
max-connect-errors = 1000000
expire-logs-days = 14
sync-binlog = 1
CACHES AND LIMITS
tmp-table-size = 16M
max-heap-table-size = 16M
query-cache-type = 0
query-cache-size = 0
max-connections = 300
thread-cache-size = 40
open-files-limit = 65535
table-definition-cache = 2048
table-open-cache = 2048
innodb-flush-method = O_DIRECT
innodb-log-files-in-group = 2
innodb-log-file-size = 32M
innodb-flush-log-at-trx-commit = 1
innodb-file-per-table = 1
innodb-buffer-pool-size = 256M
Another item that came to my attention is that the fail2ban service comes up very often and with a large memory footprint when I run the “top” command.
I wonder if there is any way to reduce the amount of memory it takes.
This is the output of the “free -m” command:
I also noticed that the average percentage of the used memory increases by 3 to 5% for each virtual server I create, even if DNS for the new domains are not pointing at my server yet and there is no traffic for these new virtual servers.
In total I am having 11 virtual servers in the system, but only 9 are used with low traffic (less than 50 visitor per day).
My biggest concern is that once one of the websites gets actual traffic that the server is unable to handle the request, because it does not seem to utilize the swap space effectively and has a high memory footprint when it is pretty much idle as it is right now.
If anyone has an idea to lower the amount of memory used or tell me that I do not need to worry, I would really appreciate it.
Thank you very much.