That question is rather wide, as there wasn't a mainframe OS but many. Equally important, headers were not made by the OS, but whatever SPOOL system was used. And there were many.
Here a typical MVS(ish) cover page:

Depending on organisational structure of OS and SPOOL-system a cover (or SPOOL) page may include:
- Job Name
- User Name
- Account Number
These three were quite often 'enlarged' for easy selection
- Process ( Task) ID
- Computer Name
- OS name
- Job Date/Time
- Job Number
- Printout Date/Time
- SPOOL File Name and Type
- Spool-ID
- Spool-In and Spool-Out Times
Some customer specific data may include items for delivery like
- Recipients Name
- Department Name
- Room Number
- Phone Number
Keep in mind, companies/institutions that had mainframes were usually not self employed in a basement, but rather large with many in house users. In early times IT departments did patch the spool system used in wired ways to modify header pages to fit their needs. In later years spool systems offered exit functions to enhance these pages in standardized ways.
It's important to keep in mind that all these are meta data not data produced in the run itself. Such will always be part of the printed data, not any cover page - although, depending on OS, they can be put on separate pages as well (as part of the content printed).
The distribution of data between a first and last page again depends on the OS, but as well on data type. Everything known ahead of printing is usually printed on the first page - which is in case of spooling is most information. The last page usually only repeats information.
To support distribution/filing most spoolers usually did put all information on the upper half of the first page (see picture), leaving the lower half for enlarged text or alike. To understand this, it's necessary to imagine how folded paper works:
- Every other page comes out face down.
On which page a printout starts is random, so 50% end up with their first page on a page-down page (*1). For distribution (*2) and filing this is not really satisfying. With all information in the upper half, printouts that started on face-down page, could have their top page folded, so all data is again visible right away.
Another way to handle this (as Another-Dave mentions) was to simply print the (first) cover page twice. Now, independent of the way the paper came out, there was always at least one cover page printed on an 'even' page.
Another detail about cover pages was that they usually started and ended with multiple lines (usually 3) of X
. This proved helpful when a printed stack was to be separated into single jobs to be delivered to its owner. It made finding cover pages easy when flipping (scanning) thru the stack. More so, with less than perfect adjusted pages these X-ed lines covered the perforation, making it visible even without flipping thru.
When high speed lasers came into use (late 70s/early 80s) this was turned into a black rectangle crossing from the cover page over to the next page, making it a definite feature to be seen right away.
So yeah, there are less obvious details of hidden in plain sight :))
*1 - Couldn't fit more page on the line :)
*2 - Usually by young ladies (or, more often grumpy old man) pushing a cart with printouts, card stacks and tapes around the building.