Calender indexing is coming in ZCS 5, but some of the code in 4.5 is queueing items for indexing. Since the server can't process the request, these operations just accumulate. The fix for this will most likely be in 4.5.11, but until then you can "reset" the redo.log like this:
* Run a full zmbackup (or wait until after the normally scheduled full backup completes)
* After it completes, stop all zimbra services
* Delete the contents of /opt/zimbra/redolog/
* Restart all services
This issue is on file as Bug 22664.
Bug 22664 - Huge number of IndexItem (type=11 or type=15) redos after server startup, never cleaned up