How to identify the cause of high iowait in Centos?

Question

linux

Zabbix is ​​running on the Centos server, there was a problem with high iowait. iostat reports %iowait =41.20. The bottleneck is most likely in the disk subsystem, because atop issues:
LVM | bbix-lv_root | busy 90% | read 32074 | write 6106e3 Tell me
how to find out the reason for loading the disk, maybe some process is loading the disk. What utilities can you find out and what actions to take in this case?

Answer 1

iotop, for example, will show the processes of their io.
With Zabbix, this happens when HouseKeeper starts working. It just tries to remove a lot of data from the history and the database starts to load the disk. If it turns out that this is the case, then set up partitioning and turn off HouseKeeper.

Answer 2

iotop.
I think you have zabbix loaded the database into the shelf, and that, in turn, is a disk. This is a well-known and long-standing problem of Zabbix.

Answer 3

iotop -oka