Calibre deals well with massive e-book collections, but it does not offer full text search (as far as I know). If you are on a Mac there is a decent search engine for your personal files built in, called Spotlight. However, Spotlight is not able to search the contents of EPUB or MOBI e-book files. This is why I decided to convert my collection to plain text.

Calibre offers the capability to bulk-convert a lot of books in parallel, but at least on my system it quite often froze after a couple of minutes. Calibre also lets you convert books from the command line with the ebook-convert command. Combined with a simple find and GNU Parallel it can be used to convert massive collections with a one liner:

find $your_library_location -type f -iname "*.mobi" | parallel --timeout 120 --progress "/Applications/calibre.app/Contents/console.app/Contents/MacOS/ebook-convert {} {.}.txt"

Parallel will try to saturate all the available CPUs on your system and can even scale out to other machines with a little more tweaking. The --timeout option will kill off any spawned sub-process that takes longer than the given ammount of seconds, which is nessecary, since sometimes the conversion seems to hang for no apparent reason. The option --progress will give you an idea how many jobs have been completed yet.

Parallel can be installed via the excellent homebrew project by running brew install parallel. The location of the ebook-convert command may vary on other systems, if I remember correctly Ubuntu sets it up correctly when installing Callibre via APT, so you can omit the path.

For me this works well enough and seems a little more robust than Calibres own job-scheduling mechanism. Hope this helps someone else.