Failed locking attempts were immediately retried up to three times
without any delay between the retries. If a lock file is not found while
checking for other locks, with the reworked backend retries there is no
delay between those retries. This is a problem if a backend requires a
few seconds to reflect file deletions in the file listings. To work
around this problem, introduce a short exponentially increasing delay
between the retries. The number of retries is now increased to 4. This
results in delays of 5, 10 and 20 seconds between the retries.
- Fix a logic error that instead of reporting the *first*
metadata-setting error that appears, we were instead reporting the
*last* error (and only if the lchown call failed!).
- Don't show any errors when setting metadata for files in non-root
mode (things like timestamps, attributes). Previously, only lchown
errors were skipped. But other kinds of attribute errors make sense
to skip as well. The code path happened to work correctly before
because of the above logic error. But once that was fixed, this
change needed to happen too.
An attempt to replace an existing file with a hardlink previously ended
with a missing file.
Remove an existing file before trying to restore a special node. This
generalizes the existing behavior for symlinks to all special node
types.
To prevent accidentally wiping all snapshots from a repository, that
option can only be used if either a snapshot filter or a keep policy is
specified.
Essentially, the option allows `forget --tag something
--unsafe-allow-remove-all` calls to remove all snapshots with a specific
tag.
`--keep-tag invalid-tag` was previously able to wipe all snapshots in a
repository. As a user specified a `--keep-*` option this is likely
unintentional. This forbid deleting all snapshot if a `--keep-*` option
was specified to prevent data loss. (Not specifying such an option
currently also causes the command to abort)
The toplevel context in restic only canceled if the user interrupts a
restic operation. If the network connection has failed this can require
waiting the full retry duration of 15 minutes which is a bad user
experience for interactive usage. Thus limit the delay to one minute in
this case.
