split files cache

[ThomasWaldmann]

Problem

the files cache can be rather large and consume a lot of memory while loaded.

currently, borg by default uses 1 files cache for all backup runs, containing informations about all files seen in the last BORG_FILES_CACHE_TTL backup runs.

there is already one mechanism to optimize that: using BORG_FILES_CACHE_SUFFIX to split this overall files cache into multiple specific files caches, so each backup run only gets what it really needs

this is a bit of a hack as the user has to care for using files cache suffixes corresponding to the backup input data sets.

From the docs:

BORG_FILES_CACHE_SUFFIX

When set to a value at least one character long, instructs borg to use a specifically named (based on the suffix) alternative files cache. This can be used to avoid loading and saving cache entries for backup sources other than the current sources.

BORG_FILES_CACHE_TTL

When set to a numeric value, this determines the maximum “time to live” for the files cache entries (default: 20). The files cache is used to quickly determine whether a file is unchanged. The FAQ explains this more detailed in: It always chunks all my files, even unchanged ones!

Ideas

auto-compute the suffix

Simply compute the cache suffix from all recursion roots, like suffix = H(root1, root2, ...)

This would use 1 files cache per borg run.

split files cache per recursion root

Maybe borg could default to using a split files cache and handle separately storing these automatically, without the user having to think about suffixes and manually mapping these to the backup runs.

E.g. it could store (and load) a files cache per recursion root.

This could potentially use multiple files caches per borg run.

Note:

• cache transactions still need to work, so we always have a correct/consistent cache.
• in some modes of operation there is no recursion root, e.g. if filenames or content data is coming in via stdin

[ThomasWaldmann]

Maybe we won't need this, see the idea in #8385.

[ThomasWaldmann]

Split by archive series (still allow manual control via env var).