It sounds like the backend scripts might have stuck in some weird states or have crashed during execution and left the config file in the /tmp directory.
Can you try to clean up all temporary files by running the following two commands on the problematic worker VMs, reboot them, and try again?
~/ioAnalyzerScripts/clean_tmp.sh
~/ioAnalyzerScripts/clean_dist.sh
If the problem still happens, please send your /var/www/ioa.log and /tmp/ioa-daemon.log files on the controller VM, and /tmp/run-iometer.log file on the failing worker VMs to io-analyzer-info (at) vmware.com so that we can collect more debug info. Also in the email, please share with us the output of `ps -ef | grep ioAnalyzer` on the failing worker VMs so that we can know whether backend scripts are still running.
Thanks,
Chien-Chia Chen