Thunderbird folder compacting changes old message headers which greatly increase size of incremental backups
Hi, TL;TR. Thunderbird should not change the whole folder (e.g. Inbox) during compacting by replacing "From" with "From - Mon May 3 20:50:30 2004" and the other way rou… (read more)
Hi,
TL;TR. Thunderbird should not change the whole folder (e.g. Inbox) during compacting by replacing "From" with "From - Mon May 3 20:50:30 2004" and the other way round. It dramatically increases the size of incremental backups done by tools detecting only blocks of bigger files which change.
Longer version. I use Borg to backup Thunderbird profile. It supports deduplication and can incrementally backup only parts of bigger files which have changed. As I use filters to automatically move messages from Inbox to the other folders, before backup, I perform compacting of the Inbox folder to effectively remove moved messages (and those deleted manually - e.g. the biggest ones). I worked pretty fine - there were no changes detected by Borg in the old messages. Only new messages were backed up.
However, it worsen in the recent months - Borg started to detected changes also in the old messages in Inbox (after compacting performed before the every backup session). As a result, a (compressed) incremental copy takes ~1/3 of the whole Inbox (instead of some MBs). For Inbox having 5GB+, it can be a problem (e.g. when - slow - network storage is used).
I diffed the Inbox file before and after of recent compacting and Thunderbird around an upgrade to 128.x (I cannot tell exactly which minor version it was) changed the "From" line for every message from just "From" to "From - Thu Nov 14 22:32:10 2024" (a date when the compaction was performed). In older backups I see that some earlier versions of Thunderbird removed the date replacing the line with just "From".
My question. Had it been an accidental change (with removing the date) which was later reverted and there will be no more "rewriting" compacted folders? Or it's "more complicated" and people using incremental backup cannot sleep peacefully?
Marcin
P.S. Let's skip the msf files for the moment as they are much smaller.