I'm trying to get size of specified directory but exclude size of one subdirectory and files inside it. Directory structure looks like this:

logs/
  archive/
    01.log
    02.log
  01.log
  02.log

I want to exclude whole 'archive' folder and all files in it. Based on documentation regex_excl is what I need.

regex_excl - regex describing the file, directory and symbolic link name pattern for exclusion (don't exclude any if empty; empty string is default value)

But using regex_excl with directory name only excludes size of directory file itself (i.e. 4096 bytes) but not whole subtree size.

Steps to reproduce:

1. Create test directory structure like described above.
2. Query size from Zabbix Agent

3. # zabbix_get -s servername -k "vfs.dir.size[/tmp/logs]"
   3347012
   # du -bcs /tmp/logs
   3347012 /tmp/logs
   3347012 total
   
   # zabbix_get -s servername -k "vfs.dir.size[/tmp/logs,,archive]"
   3342916
   # du --exclude=archive -bcs /tmp/logs
   1850526 /tmp/logs
   1850526 total
   
   # zabbix_get -s servername -k "vfs.dir.size[/tmp/logs,,\.log]"
   8192
   # du --exclude=*log -bcs /tmp/logs
   8192    /tmp/logs
   8192    total

Result:

• With first request we get size of all files in target folder (/tmp/logs). du return same size.
• With second request we get size of all files excluding size of 'archive' directory file (3347012 - 3342916 = 4096 bytes). 'du' returns different result because it does not count size of files in 'archive' directory.****
• 3-rd request returns only size of 2 directory files: 'logs' and 'archive' all '*.log' files are excluded.

Expected:
Based on documentation description I expect regex_excl operate same way as --exclude key of du utility, but it behaves in different, unpredictable way.