[prev in list] [next in list] [prev in thread] [next in thread] 

List:       debian-user
Subject:    Re: partition reporting full, but not
From:       DdB <debianlist () potentially-spam ! de-bruyn ! de>
Date:       2024-02-19 10:58:19
Message-ID: a1b78b27-580c-a971-959c-3700861652aa () potentially-spam ! de-bruyn ! de
[Download RAW message or body]

Am 19.02.2024 um 04:20 schrieb Keith Bainbridge:
> I am convinced that the missing space is used by btrfs snapshot process.

First off: I am not a btrfs user (and will never be, i might add).
I am using zfs since many years, and - although i read an awful lot of
documentation beforehand, and played around with it quite a bit, i had
to learn a couple of lessons "the hard way".

To me, what i am inferring from your mails, are a couple of conceptual
misundestandings of what a COW-filesystem is, and how to best make use
of it. Especially the time-machine using snapshots and hardlinks seems
very kinky to me.

I am using zfs somewhat in a similar way, but it took me some years to
set it up in a good way. Apart from a warning and the usual RTFM, i do
not have much to offer. Except maybe for this: Snapshots hold space,
they are not like hardlinks. (although one can snapshot hardlinks ;-) )

Once, some older filesystem state is snapshotted, its space is taken
until the snapshot gets destroyed/removed. This can lead to situations,
where a device is full, but in order to free some space, deleting files
will NOT help, because in order to do so, a change in the directory
needs to be made, but there may not be the room to create a copy of it
in the first place.
And deleting files from a snapshot is not feasable (at least not in zfs).

To my eyes, your handling of the system looks somewhat risky and not
well planned. You may go on this way, as long as you consider the whole
thing as being part of a longish learning session. But do not trust
important data being safe this way!

just my 2 cents

[prev in list] [next in list] [prev in thread] [next in thread]