Is this a bugfix, or an enhancement? Please describe.
Enhancement.

After a recent hardware upgrade and mandated security software upgrade on our servers, we have noticed that using enable-sync=true on the rundeck localhost node is causing massive performance issues. For example. we have a job that runs across around 20 nodes takes around 6 minutes with enable-sync=false, but takes around an hour with enable-sync=true.

This change keeps the enable-sync=true behaviour that fixes the "text file busy" issue we were seeing, but keeps the performance on par with enable-sync=false.

see#9544

Describe the solution you've implemented
sync is now called with filepath so that it only syncs that specific file of all outstanding changes, see the sync docs for more info:

https://man7.org/linux/man-pages/man1/sync.1.html

Without the filepath, the sync command synchronizes all cached data for the current user to the permanent memory.

Describe alternatives you've considered
see rundeck google group discussion:
https://groups.google.com/g/rundeck-discuss/c/Eqetx33hKgM

I have been testing with enable-sync=false and file-busy-err-retry=true in hopes that this would prevent the text file busy issue while also disabling the sync command. This did not work.

I also tried setting
file-copy-destination-dir="/var/lib/rundeck/rundeckTmpfs"

and mounting a tmpfs filesystem.

I would have thought setting this along with enable-sync=false could workaround our problem, as the tmpfs location would be in memory and the sync shouldn't be needed/have no effect. This also didn't fix the issue. We still see intermittent text file busy failures.