Spent all day yesterday trying to figure out why a client's logs weren't indexing. Most of the time I had no access to the server in question, so I was simply troubleshooting from internal logs, configs, and the sporadic logs that would show up briefly after a restart.
Finally, when I was just about to throw in the towel, I started poking around directories above the target files. The monitor line had an asterisk at this point in the path, so even though most other dirs didn't match further down the line, Splunk had to check them. Several of them had 100k+ files in them. So Splunk was stuck trying to read these dirs. Even performing just an ls | wc -l took over 10 minutes on a few of them.
I can find big directories with something like this and send it into Splunk for alerting:
find /path -size +100k -type d
Adjusting the size requirement as needed. Is there a better way to avoid these landmines?