Large 000000.vlog file size

[luedi0]

Over two years of runtime, the vlog file has grown to 100MB. Parsing this file is taking up considerable resources. Is there a way of pruning the data without losing functionality? Perhaps removing old entries?

[claude]

Unfortunately, there is currently no built-in mechanism to prune or compact the database. This is a known limitation of the current implementation.

A few notes on how the database works:

• Episode records are never physically removed from the database, even when keep_last is configured — that option only deletes the downloaded files, not the database entries
• Episodes that are cleaned up have their status changed to EpisodeCleaned but remain in the DB indefinitely
• The only records ever removed are episodes that disappear from the platform API and were never downloaded (status EpisodeNew)

The [database.badger] section does expose a couple of BadgerDB options, but neither helps with data pruning:

[database.badger]
truncate = true   # recover from corrupted value log (not for size reduction)
file_io = true    # use file I/O instead of mmap

Adding a database compaction/garbage collection option would require a new feature. This issue is a good candidate for tracking that enhancement. One potential approach would be a CLI flag or config option to prune EpisodeCleaned records older than a certain age from the database.