I've checked the script and was surprised how readable & well documented it is. All seems to be fine & it even correctly handles multiple store databases (needed due to the limited single-file-size on FAT32).

A couple of possibilities what might be causing the missing tiles:

• old tiles - modRana previously didn't check downloaded tiles, so sometimes some tiles might actually have been error pages - but if the tiles were showing fine in offline mode and now aren't, this can't be it
• bug in modRana - modRana fails to load the tiles from the database for some reason
• bug in the import script - I'm not that well versed in perl, so maybe there is still some edge-case that is intermittently triggered

Anyway, I'm currently considering finally adding the tile import functionality to modRana - probably first through the command line interface, followed by support in the GUIs.
The tile storage logic is all there, it just needs to know from where to take the tiles and to which database they should be added.

About the modRana data repository
(http://data.modrana.org/monav/)

First, I would like to thank the Natural Language Processing Centre of the Masaryk University for providing the hosting & computational resources needed to generate the data in the first place. Thanks a lot !

Coverage
The whole world should be covered, 2-3x times. There are data packs for individual countries, continents and for some countries (Germany, France, Canada,...) also for local regions. There are separate car, bike & pedestrian data packs.

BTW, the car routing data size for Europe is about 3 GB, so it should fit on the MyDocs FAT32 partition without issues.

Installing data for modRana
Installation is the same as with previous data packs - they are just in tar.gz, not in zip. The car/bike/pedestrian pack include the shared data structur, just unpack them to the usual /home/user/MyDocs/.maps/monav_data/ directory.

Example (on the N900):

Statistics
Total repository size: 91 GB
Pack count: about 265
Source data downloads: 37.18GB source OSM data in PBF format
Total run length: about 5 hours on a machine with 48 CPU cores (including all stages - download, processing & packaging)

Software
The modRana data repository software, used to generate the repository, is hosted on Github:
https://github.com/M4rtinK/modrana-data-repository
If you want to have your own repository, just run repository.py and the repository will be created inside the results directory.

I plan to write a proper post about how the repository software works & I'm also still working on it, mainly on more configuration options & providing better feedback while running a repository update.

Planed features

• repository manifest - so that the repository can be used for downloading data directly from modRana
• *.zsync files for incremental data updates
• regular repository updates - about every 3 days
• maybe also MD5SUMS ?