The previous approach of rewriting all snapshots first, then flushing
the repository data and finally removing old snapshots has the downside
that an interrupted command execution leaves behind broken snapshots as
not all new data is already flushed.
This adds support for caching already rewritten trees, handling of load
errors and disabling the check that the serialization doesn't lead to
data loss.
The more generic RewriteNode callback replaces the SelectByName and
PrintExclude functions. The main part of this change is a preparation to
allow using the TreeRewriter for the `repair snapshots` command.
The files in a tree must be sorted in lexical order. However, this
cannot be guaranteed when appending a filename suffix. For two files
file, file.rep
where "file" is broken, this would result in
file.repaired, file.rep
which is no longer sorted.
In addition, adding a filename suffix is also prone to filename
collisions which would require a rather complex search for a
collision-free name in order to work reliably.
Simplify CLI options:
* Rename "DeleteSnapshots" to "Forget"
* Replace "AddTag" and "Append" with hardcoded values
Change output and snapshot modifications to be more in line with the
"rewrite" command.
The builtin mechanism to capture a stacktrace in Go is to send a SIGQUIT
to the running process. However, this mechanism is not avaiable on
Windows. Thus, tweak the SIGINT handler to dump a stacktrace if the
environment variable `RESTIC_DEBUG_STACKTRACE_SIGINT` is set.
When saving files to the local backend, in some cases the used fsync
calls are slow enough to cause the tests to time out. Thus, increase the
test timeouts as a stopgap measure until we can use the mem backend for
these tests.
The SemaphoreBackend now uniformly enforces the limit of concurrent
backend operations. In addition, it unifies the parameter validation.
The List() methods no longer uses a semaphore. Restic already never runs
multiple list operations in parallel.
By managing the semaphore in a wrapper backend, the sections that hold a
semaphore token grow slightly. However, the main bottleneck is IO, so
this shouldn't make much of a difference.
The key insight that enables the SemaphoreBackend is that all of the
complex semaphore handling in `openReader()` still happens within the
original call to `Load()`. Thus, getting and releasing the semaphore
tokens can be refactored to happen directly in `Load()`. This eliminates
the need for wrapping the reader in `openReader()` to release the token.
This turns snapshotFilterOptions from cmd into a restic.SnapshotFilter
type and makes restic.FindFilteredSnapshot and FindFilteredSnapshots
methods on that type. This fixes #4211 by ensuring that hosts and paths
are named struct fields instead of unnamed function arguments in long
lists of such.
Timestamp limits are also included in the new type. To avoid too much
pointer handling, the convention is that time zero means no limit.
That's January 1st, year 1, 00:00 UTC, which is so unlikely a date that
we can sacrifice it for simpler code.
The output is now
```
-v, --verbose be verbose (specify multiple times or a level using --verbose=n, max level/times is 2)
```
instead of
```
-v, --verbose n be verbose (specify multiple times or a level using --verbose=n, max level/times is 2)
```
Fixes restic#719
If the option is passed, restic will wait the specified duration of time
and retry locking the repo every 10 seconds (or more often if the total
timeout is relatively small).
- Play nice with json output
- Reduce wait time in lock tests
- Rework timeout last attempt
- Reduce test wait time to 0.1s
- Use exponential back off for the retry lock
- Don't pass gopts to lockRepo functions
- Use global variable for retry sleep setup
- Exit retry lock on cancel
- Better wording for flag help
- Reorder debug statement
- Refactor tests
- Lower max sleep time to 1m
- Test that we cancel/timeout in time
- Use non blocking sleep function
- Refactor into minDuration func
Co-authored-by: Julian Brost <julian@0x4a42.net>
This turns snapshotFilterOptions from cmd into a restic.SnapshotFilter
type and makes restic.FindFilteredSnapshot and FindFilteredSnapshots
methods on that type. This fixes #4211 by ensuring that hosts and paths
are named struct fields instead of unnamed function arguments in long
lists of such.
Timestamp limits are also included in the new type. To avoid too much
pointer handling, the convention is that time zero means no limit.
That's January 1st, year 1, 00:00 UTC, which is so unlikely a date that
we can sacrifice it for simpler code.
The output is now
```
-v, --verbose be verbose (specify multiple times or a level using --verbose=n, max level/times is 2)
```
instead of
```
-v, --verbose n be verbose (specify multiple times or a level using --verbose=n, max level/times is 2)
```
This method had a buffer argument, but that was nil at all call sites.
That's removed, and instead LoadUnpacked now reuses whatever it
allocates inside its retry loop.
The StdioWrapper is not used at all by the ProgressPrinters. It is
called a bit earlier than previously. However, as the password prompt
directly accessed stdin/stdout this doesn't cause problems.
The maximum for `--verbose=n` is n=2. Internally it is translated into a
scale from 0 to 3. However, the default (without verbose) is 1, thus the
verbosity level can only be increased two times.
Only the repacking of *un*compressed packs reduces the amount of
uncompressed data. Previously the counter even overflowed for fully
compressed repositories.
Commands should use the normal shutdown path. In addition, the Exitf
function was only used by `dump` and `restore` but not any other command
which introduces the risk of inconsistent behavior.
When reporting an error for a tree, the output message can overlap with
the progress bar output, e.g. `error for tree e91ef6fb:napshots`.
The fix only applies for this specific message and does not work on
Windows.
Revert what seems to be a typo introduced as part of the fix for #2041
in 2018 7d0f2eaf24.
`xbuild` does not look like a go build/tag keyword to me, I failed to
find documentation for it and using `go install -tags '!selfupdate' ...`
has no effect, i.e. self-update code is still compiled.
`+build` however works; updating the OpenBSD port/binary package
security/restic to apply this PR works as expected:
```
$ restic help | grep self
$ restic self-update
unknown command "self-update" for "restic"
```
(Using `go:build` now as per restic's style and gofmt.)
Previously, using `restic-0.14.0p1` on OpenBSD/amd64 7.2-current would
check for a newer version and probably attempt replacing the system wide
root-owned executable (on a read-only filesystem) as unprivileged user:
```
$ restic version
restic 0.14.0 compiled with go1.19.2 on openbsd/amd64
$ restic help | grep self
self-update Update the restic binary
$ restic self-update
writing restic to /usr/local/bin/restic
find latest release of restic at GitHub
restic is up to date
```
(It never tried to actually write besaid path; doing so would fail, so
the current message can be considered misleading.)
The scanner process has only cosmetic effect for the progress printer,
and can be disabled without impacting functionality when the user does
not need an estimate of completion.
In many cases the scanner process can provide beneficial priming of
the file system cache, so as general advice it should not be disabled.
However, tests have shown that backup of NFS and fuse based filesystems,
where stat(2) is relatively expensive, can be significantly faster
without the scanner.
The Original field is meant to remember the original snapshot id if e.g.
changing its tags. It was only set by the `rewrite` command if it was
not set previously. However, a rewritten snapshot is potentially rather
different from the original snapshot. Thus just always set the Original
field. This also makes it easier to later on detect and potentially
remove the original snapshots.
