index=* | eval delay_secs=abs(_time - _indextime) | where delay_secs > 1800 | eval delay_mins=(delay_secs / 60) | eval delay_hrs=(delay_mins/60) | stats count avg(delay_mins) avg(delay_hrs) by host,source
-todd
splunk> I am speed
Todd "shamgow" Gow
Sr. Inside Sales Engineer
Splunk, Inc.
Office: +1.972.244.8753
Mobile: +1.972.955.2552
Email: tg...@splunk.com
Plano > San Francisco Cupertino London Hong Kong Washington D.C. Seattle

This message is intended only for the personal, confidential, and authorized use of the recipient(s) named above. If you are not that person, you are not authorized to review, use, copy, forward, distribute or otherwise disclose the information contained in the message.
Here is the actual search I have been using to find TZ/clock problems:
index=* | eval delay_secs=abs(_time - _indextime) | where delay_secs > 1800 | eval delay_mins=(delay_secs / 60) | eval delay_hrs=(delay_mins/60) | stats count avg(delay_mins) avg(delay_hrs) by host,source
I ran this on my test system and am finding very "interesting" results. I am not totally clear what I do with this info.