fixes commit b4fb262335, which copied
over the "Removed bad files:" block, but only adapted the log text,
not the actual variable.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
saves files mtime as i64 instead of u64 which enables backup of
files with negative mtime
the catalog_decode_i64 is compatible to encoded u64 values (if < 2^63)
but not reverse, so all "old" catalogs can be read with the new
decoder, but catalogs that contain negative mtimes will decode wrongly
on older clients
also remove the arbitrary maximum value of 2^63 - 1 for
encode_u64 (we just use up to 10 bytes now) and correctly
decode them and update the comments accordingly
adds also test for i64 encode/decode and for compatibility between
u64 encode and i64 decode
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
commit a4915dfc2b made a wrong fix, as
it did not observed that the last expressions was done under the
invariant that we had a last verification result, because if none
could be loaded we already returned true (include).
It thus broke the case for "never re-verify", which is important when
using multiple schedules, a more high frequent one for new,
unverified snapshots, and a low frequency to re-verify older snapshots,
e.g., monthly.
Fix this case again, rework the code to avoid this easy to oversee
invariant. Use a nested match to better express the implication of
each setting, and add some comments.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
and load it again when opening it
this way we can persist the status of the last garbage collect across
daemon reloads and reboots
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
To cater to the paranoid, a new datastore-wide setting "verify-new" is
introduced. When set, a verify job will be spawned right after a new
backup is added to the store (only verifying the added snapshot).
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Force consumers to use the lookup_datastore method instead of
potentially opening a datastore twice, and pass the config we have
already loaded into open_with_path, removing the need for open(1).
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Avoid races when updating manifest data by flocking a lock file.
update_manifest is used to ensure updates always happen with the lock
held.
Snapshot deletion also acquires the lock, so it cannot interfere with an
outstanding manifest write.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Removing a snapshot has some more safety checks which we don't want to
ignore when removing an entire group (i.e. locking the manifest and
notifying GC).
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
There's no point in having that as a seperate method, just parse the
thing into a struct and write it back out correctly.
Also makes further changes to the method simpler.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
If we can't acquire a lock (either because the snapshot disappeared, it
is about to be forgotten/pruned, or it is currently still running) skip
the snapshot. Hold the lock during verification, so that it cannot be
deleted while we are still verifying.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
...to avoid it being forgotten or pruned while in use.
Update lock error message for deletions to be consistent.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
To untangle the server code from the actual backup
implementation.
It would be ideal if the whole backup/ dir could become its
own crate with minimal dependencies, certainly without
depending on the actual api server. That would then also be
used more easily to create forensic tools for all the data
file types we have in the backup repositories.
Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
This is only acquired in those two methods, both as shared. So it has
no use.
It seems, that it was planned in the past that the index deletion
should take the exclusive, while read and write takes the shared
flock on the index, as one can guess from the lock comments in commit
0465218953
But then later, in commit c8ec450e37)
the documented semantics where changed to use a temp file and do an
atomic rename instead for atomicity.
The reader shared flock on the index file was done inbetween,
probably as preparatory step, but was not removed again when strategy
was changed to using the file rename instead.
Do so now, to avoid confusion of readers and a useless flock.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
In theory, one can do std::mem::forget, and ignore the drop handler. With
the lifetime hack, this could result in a crash.
So we simply require 'static lifetime now (futures also needs that).
- remove chrono dependency
- depend on proxmox 0.3.8
- remove epoch_now, epoch_now_u64 and epoch_now_f64
- remove tm_editor (moved to proxmox crate)
- use new helpers from proxmox 0.3.8
* epoch_i64 and epoch_f64
* parse_rfc3339
* epoch_to_rfc3339_utc
* strftime_local
- BackupDir changes:
* store epoch and rfc3339 string instead of DateTime
* backup_time_to_string now return a Result
* remove unnecessary TryFrom<(BackupGroup, i64)> for BackupDir
- DynamicIndexHeader: change ctime to i64
- FixedIndexHeader: change ctime to i64
since converting from i64 epoch timestamp to DateTime is not always
possible. previously, passing invalid backup-time from client to server
(or vice-versa) panicked the corresponding tokio task. now we get proper
error messages including the invalid timestamp.
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
by either printing the original, out-of-range timestamp as-is, or
bailing with a proper error message instead of panicking.
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
even if it can't be handled by chrono. silently replacing it with epoch
0 is confusing..
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
We need to update the atime of chunk files if they already exist,
otherwise a concurrently running GC could sweep them away.
This is protected with ChunkStore.mutex, so the fstat/unlink does not
race with touching.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
The iterator of get_chunk_iterator is extended with a third parameter
indicating whether the current file is a chunk (false) or a .bad file
(true).
Count their sizes to the total of removed bytes, since it also frees
disk space.
.bad files are only deleted if the corresponding chunk exists, i.e. has
been rewritten. Otherwise we might delete data only marked bad because
of transient errors.
While at it, also clean up and use nix::unistd::unlinkat instead of
unsafe libc calls.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
This ensures that following backups will always upload the chunk,
thereby replacing it with a correct version again.
Format for renaming is <digest>.<counter>.bad where <counter> is used if
a chunk is found to be bad again before a GC cleans it up.
Care has been taken to deliberately only rename a chunk in conditions
where it is guaranteed to be an error in the chunk itself. Otherwise a
broken index file could lead to an unwanted mass-rename of chunks.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Save the state ("ok" or "failed") and the UPID of the respective
verify task. With this we can easily allow to open the relevant task
log and show when the last verify happened.
As we already load the manifest when listing the snapshots, just add
it there directly.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
to avoid `map_struct` which is actually unsafe because it
does not verify alignment constraints at all
Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
This reverts commit d53fbe2474.
The HashSet and "register" function are unnecessary, as we already know
which backup is the one we need to check: the last one, stored as
'last_backup'.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
An flock on the snapshot dir itself is used in addition to the group dir
lock. The lock is used to avoid races with forget and prune, while
having more granularity than the group lock (i.e. the group lock is
necessary to prevent more than one backup per group, but the snapshot
lock still allows backups unrelated to the currently running to be
forgotten/pruned).
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Attempt to lock the backup directory to be deleted, if it works keep the
lock until the deletion is complete. This way we ensure that no other
locking operation (e.g. using a snapshot as base for another backup) can
happen concurrently.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
an encrypted Index should never reference a plain-text chunk, and an
unencrypted Index should never reference an encrypted chunk.
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
If the datastore holds broken backups for some reason, do not attempt to
base following snapshots on those. This would lead to an error on
/previous, leaving the client no choice but to upload all chunks, even
though there might be potential for incremental savings.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
instead of bailing and stopping the entire GC process, warn about the
missing chunks and continue.
this results in "TASK WARNINGS: X" as the status.
Signed-off-by: Oguz Bektas <o.bektas@proxmox.com>
Used chunks are marked in phase1 of the garbage collection process by
using the atime property. Each used chunk gets touched so that the atime
gets updated (if older than 24h, see relatime).
Should there ever be a situation in which the phase1 in the GC run needs
a very long time to finish, it could happen that the grace period
calculated in phase2 is not long enough and thus the marking of the
chunks (atime) becomes invalid. This would result in the removal of
needed chunks.
Even though the likelyhood of this happening is very low, using the
timestamp from right before phase1 is started, to calculate the grace
period in phase2 should avoid this situation.
Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
regular chunks are only decoded when their contents are accessed, in
which case we need to have the key anyway and want to verify the digest.
for blobs we need to verify beforehand, since their checksums are always
calculated based on their raw content, and stored in the manifest.
manifests are also stored as blobs, but don't have a digest in the
traditional sense (they might have a signature covering parts of their
contents, but that is verified already when loading the manifest).
this commit does not cover pull/sync code which copies blobs and chunks
as-is without decoding them.
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
Errors while applying metadata will not be considered fatal
by default using `pxar extract` unless `--strict` was passed
in which case it'll bail out immediately.
It'll still return an error exit status if something had
failed along the way.
Note that most other errors will still cause it to bail out
(eg. errors creating files, or I/O errors while writing
the contents).
Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
so that we can print a list at the end of the worker which backups
are corrupt.
this is useful if there are many snapshots and some in between had an
error. Before this patch, the task log simply says to 'look in the logs'
but if the log is very long it makes it hard to see what exactly failed.
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
this makes it easier to see which chunks are corrupt
(and enables us in the future to build a 'complete' list of
corrupt chunks)
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
This should never trigger if everything else works correctly, but it is
still a very cheap check to avoid wrongly marking a backup as "OK" when
in fact some chunks might be missing.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Multiple backups within one backup group don't really make sense, but
break all sorts of guarantees (e.g. a second backup started after a
first would use a "known-chunks" list from the previous unfinished one,
which would be empty - but using the list from the last finished one is
not a fix either, as that one could be deleted or pruned once the first
simultaneous backup is finished).
Fix it by only allowing one backup per backup group at one time. This is
done via a flock on the backup group directory, thus remaining intact
even after a reload.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
To prevent a race with a background GC operation, do not allow deletion
of backups who's index might currently be referenced as the "known chunk
list" for successive backups. Otherwise the GC could delete chunks it
thinks are no longer referenced, while at the same time telling the
client that it doesn't need to upload said chunks because they already
exist.
Additionally, prevent deletion of whole backup groups, if there are
snapshots contained that appear to be currently in-progress. This is
currently unlikely to trigger, as that function is only used for sync
jobs, but it's a useful safeguard either way.
Deleting a single snapshot has a 'force' parameter, which is necessary
to allow deleting incomplete snapshots on an aborted backup. Pruning
also sets force=true to avoid the check, since it calculates which
snapshots to keep on its own.
To avoid code duplication, the is_finished method is factored out.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
And make verify_crc private for now. We always call load_from_reader() to
verify the CRC.
Also add load_chunk() to datastore.rs (from chunk_store::read_chunk())
useful to get info like, was the previous snapshot encrypted in
libproxmox-backup-qemu
Requested-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Requires updating the AsyncRead implementation to cope with byte-wise
seeks to intra-chunk positions.
Uses chunk_from_offset to get locations within chunks, but tries to
avoid it for sequential read to not reduce performance from before.
AsyncSeek needs to use the temporary seek_to_pos to avoid changing the
position in case an invalid seek is given and it needs to error in
poll_complete.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
