I have a url trigger with a text context and default regex of ".*" The cron schedule is "0 08-16 * * 1-5", which is to say business hours during business days.

Each morning the trigger runs at 8am to check the content and seems to have forgotten the state from the last invocation at 4pm the previous day. It prints:

Inspecting TXT content for URL https://jenkins/job/myjob/lastSuccessfulBuild/buildNumber

Polling started on Mar 19, 2014 8:01:19 AM
Polling for the job myjob
Looking nodes where the poll can be run.
Looking for a node to the restricted label master.
Restrict on master label. Polling on master.

Polling on master.
Using Basic Authentication with the user 'user'
Invoking the url:
https://jenkins/job/myjob/lastSuccessfulBuild/buildNumber
Inspecting the content
Capturing URL context. Waiting next schedule to check a change.

Polling complete. Took 0.66 sec.
No changes.

This effectively means the earliest the job can be triggered is the second cron trigger of the day. I would expect that the URL Trigger plugin would remember the value from the previous day and calculate changes from that. Jenkins has not been restarted overnight.