Description of problem:
Today I was wondering why many of our new installed systems has a load not lower than 1. I found that all have a "find" process running doing high io:

find . -daystart -name NoSuchFile

The process has a cwd of "/", this means it searches every directory in the tree, even all nfs mounted directories! In our environment this creates a useless unnecessary high load on our nfs servers, because many nfs clients run the same find command at the same time all days!

This are the process tree that call the find process:

systemd --system --deserialize 55
  └─pmlogger /usr/share/pcp/lib/pmlogger start
      └─pmlogger_check /usr/libexec/pcp/bin/pmlogger_check -V
          └─pmlogger_daily /usr/libexec/pcp/bin/pmlogger_daily -K -V
              └─find . -daystart -name NoSuchFile

/usr/libexec/pcp/bin/pmlogger_daily contains the line
find . -daystart -name NoSuchFile >/dev/null 2>&1 && FIND_DAYSTART="-daystart"

Who writes such horrible code and releases it to the world?

This line is new in pcp 4.3.0:
https://github.com/performancecopilot/pcp/commit/7570a314ed657589f31830dc880d811b6ff589e9

Please do not run find every day on / without restricting to a) local filesystems, b) a small subdirectory! It creates a unnecessary load to systems that reduces the performance for the users, and that stresses the disks (and in this case the network and nfs servers too).



Version-Release number of selected component (if applicable):
pcp-4.3.0-1.fc29.x86_64

How reproducible:
Always.

Steps to Reproduce:
1. Install (current version of) pcp.
2. Wait.

Actual results:
Extra load because of long running find command.

Expected results:
No extra load.