Re: [Errno 28] No space left on device

[Dominic Raferd]

On Thu, 14 Nov 2019 at 19:12, <address@hidden> wrote:

> Hi,
>
> On 14/11/2019 01:47, Yves Bellefeuille wrote:
> >> 3. which file system type you are using to backup?
> > Backing-up from NTFS (Windows 7) to ext4.
>
> Cross-filesystem-types backup is always a bit tricky. This said if you
> have the occasion, it would be nice to test the latest version from
> master. We could use some real life testers.
>
> >
> >> 4. do you also get the issue if you use the `--no-eas` option?
> > Indeed, this seems to have solved the problem.
> >
> > I wonder why this suddenly failed after working for several years.
> > Thanks for the help.
>
> Here I can possibly offer you an explanation: according to man 5 xattr,
> the ext4 file system has a limit of extended attributes size (1, 2 or
> 4KB), and you might have now a Windows file with bigger EAs, which leads
> to the issue:
>
> touch xxx
> python3
>  >>> x = '0123456789' * 1024  # 10KB size
>  >>> import xattr
>  >>> xattr.setxattr('xxx', 'user.comment', x)
> Traceback (most recent call last):
>    File "<stdin>", line 1, in <module>
> OSError: [Errno 28] No space left on device
>
> So you have multiple solutions:
>
> 1. switch from ext4 to XFS, which doesn't know this limit (I checked)
> 2. identify the culprit file and trim the EAs (somehow?!?)
> 3. use the --no-eas option, not knowing what you lose as information
>
> I created an issue against pyxattr:
> https://github.com/iustin/pyxattr/issues/23


Unexpected out of space errors often (or even usually) relate to the
temporary file location. Try increasing the space in this location or
specifying a different one using rdiff-backup's --tempdir or
--remote-tempdir options. Regressions (which are attempted automatically by
rdiff-backup if it finds a damaged repository) are especially
temp-intensive. I have a 60GB SSD entirely dedicated to /tmp.