The one who eats my iotop or "where to write logs"?

Question

linux

We have a regular HDD iotop -o -ain ½ hour

USER	DISK WRITE	IO	COMMAND
root	54,52 M		0,00%	sw-collectd -C /etc/sw-collectd/collectd.conf
        
root	77,27 M		0,71%	[flush-253:0]
root	49,90 M		1,74%	[jbd2/dm-0-8]
        
apache	19,04 M		0,01%	httpd
apache	18,16 M		0,04%	httpd
apache	14,61 M		0,02%	httpd
apache	12,80 M		0,01%	httpd
apache	12,63 M		0,03%	httpd
apache	12,11 M		0,02%	httpd
apache	10,42 M		0,02%	httpd
apache	5,02 M		0,01%	httpd
apache	4,83 M		0,01%	httpd
nginx	4,72 M		0,01%	nginx worker
        
mysql	2,37 M		0,02%	mysqld
mysql	2,03 M		0,00%	mysqld
mysql	88,00 K		0,00%	mysqld

Summary:
* sw-collectd is Plesk's legacy. Some kind of gluttonous recording (but not IO) monitoring.
* flush-253:0 + jbd2/dm-0-8 - disk logging (?).
* httpd/nginx - and this is the main source of disk recording - LOGS!
It doesn't seem to be a big problem. Well, it takes ± 500Gb for a month on the disk system, well, you eat off this 0.2% IO, but you want to understand what you should pay attention to today, so that there would be fewer questions tomorrow.
Is there any administrator Zen on the disk theme?
Should the logs of the last hour be transferred to a non-journaled system?
Simplify the logs and catch code errors so that they don’t crash into the log?

Answer 1

You can compress logs before writing: nginx.org/r/access_log/ru

If the gzip option is given, then the buffer will be compressed before being written to the file.

Typical logs are 7-15 times compressed on average with gzip.

Answer 2

Well, the situation is simple: either you need logs or you don't. If you are not interested in access, then just disable it.