Commit Graph

2124 Commits

Author SHA1 Message Date
Michael Eischer 5e0ea8fcfa pack: move to repository package 2024-05-25 13:13:03 +02:00
Michael Eischer 50ec408302 index: move to repository package 2024-05-25 13:13:03 +02:00
Michael Eischer 8e5d7d719c cache: move to backend package 2024-05-24 23:04:06 +02:00
Michael Eischer 3c7b7efdc9 repository: remove prune plan parts once they are no longer necessary 2024-05-24 22:18:14 +02:00
Michael Eischer 462b82a060 index: reduce size of compressed indexes
use the same index size for compressed and uncompressed indexes.
Otherwise, decoding the index of a compressed repository requires
significantly more memory.
2024-05-24 22:18:14 +02:00
Michael Eischer 77873f5a9d repository: let prune control data structure of usedBlobs set 2024-05-24 22:18:14 +02:00
Michael Eischer 2033c02b09 index: replace CountedBlobSet with AssociatedSet 2024-05-24 22:18:14 +02:00
Michael Eischer 93098e9265 prune: hide implementation details of counted blob set 2024-05-24 21:42:56 +02:00
Michael Eischer 860b595a8b backend: increase watchdog test timeout for deflaking 2024-05-24 21:33:17 +02:00
Michael Eischer 027cc64737 repository: fix prune heuristic to allow resuming interrupted runs
Pack files created by interrupted prune runs, appear to consist only of
duplicate blobs on the next run. This caused the previous heuristic to
ignore those pack files. Now, a duplicate blob in a specific pack file
is also selected if that pack file only contains duplicate blobs. This
allows prune to select the already rewritten pack files.
2024-05-24 21:33:17 +02:00
Michael Eischer e52033a8bd index: slightly reduce Rewrite concurrency
The index operations are likely CPU-bounded. Thus, reduce the
concurrency accordingly.
2024-05-24 21:33:17 +02:00
Michael Eischer 57d69aa640 index: cleanup SaveIndex method 2024-05-24 21:33:17 +02:00
Michael Eischer 2ca1c37721 index: additional tests for new index save methods 2024-05-24 21:33:17 +02:00
Michael Eischer 5f7b48e65f index: replace Save() method with Rewrite and SaveFallback
Rewrite implements a streaming rewrite of the index that excludes the
given packs. For this it loads all index files from the repository and
only modifies those that require changes. This will reduce the index
churn when running prune. Rewrite does not require the in-memory index
and thus can drop it to significantly reduce the memory usage.

However, `prune --unsafe-recovery` cannot use this strategy and requires
a separate method to save the whole in-memory index. This is now handled
using SaveFallback.
2024-05-24 21:33:17 +02:00
Michael Eischer ad98fbf7dd restic: add IDSet.Clone() method 2024-05-24 21:33:17 +02:00
Michael Eischer 72482ce5bd index: misc cleanups 2024-05-24 21:33:17 +02:00
Michael Eischer 68fa0e0305 prune: no longer disable automatic index updates
this allows prune to resume an interrupted prune run.
2024-05-24 21:33:17 +02:00
Michael Eischer 9aa0c90fb2 index: remove supersedes field
Using the field with its current semantics is nearly impossible to get
right. Remove it as it will be replaced anyways in repository format 3.
2024-05-24 21:33:17 +02:00
Michael Eischer 76e6719f2e repository: make CreateIndexFromPacks method private 2024-05-24 21:33:17 +02:00
Michael Eischer 04ad9f0c0c repository: remove Packer and SavePacker from public interface 2024-05-24 21:33:17 +02:00
Michael Eischer 550d1eeac3 repository: remove SaveIndex from interface
The method is now only indirectly accessible via Prune or RepairIndex.
2024-05-24 21:33:17 +02:00
Michael Eischer fb59e00614 index: rewrite MasterIndex load/save test to be independent of repository 2024-05-24 21:33:17 +02:00
Michael Eischer 447b486c20 index: deduplicate index loading of check and repository 2024-05-24 21:33:17 +02:00
Michael Eischer 6ca12c1b4a archiver: replace most uses of restic.Repository 2024-05-24 21:33:17 +02:00
Michael Eischer 864995271e repository: unwrap BlobHandle parameters of LookupBlob
The method now uses the same parameters as LookupBlobSize.
2024-05-24 21:33:17 +02:00
Michael Eischer 1266a4932f repository: fix parameter order of LookupBlobSize
All methods should use blobType followed by ID.
2024-05-24 21:33:17 +02:00
Michael Eischer e848ad651a restic: name parameters in restic interface 2024-05-24 21:33:17 +02:00
Michael Eischer 0bb0720348 test cleanups 2024-05-24 21:33:17 +02:00
Michael Eischer c01bcb1001 archiver: remove unused masterIndex from test 2024-05-24 21:33:17 +02:00
Michael Eischer 0aa5c53842 repository: replace HasBlob with LookupBlobSize 2024-05-24 21:33:17 +02:00
Michael Eischer 8f1e70cd9b repository: remove clearIndex and packSize from public interface 2024-05-24 21:33:17 +02:00
Michael Eischer 4df887406f repository: inline MasterIndex interface into Repository interface 2024-05-24 21:33:17 +02:00
Michael Eischer 3cc6827f09 forget: add feature flag for safe --keep-tags behavior 2024-05-24 20:45:33 +02:00
Michael Eischer 57f9739573 forget: Add --unsafe-allow-remove-all option
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.
2024-05-24 20:45:33 +02:00
Michael Eischer c0e1f36830 forget: refuse deleting the last snapshot in a snapshot group
`--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)
2024-05-24 20:45:33 +02:00
Michael Eischer d106ad6921 restic: regenerate snapshot keep policy golden test files 2024-05-24 20:45:33 +02:00
Michael Eischer e4a48085ae backend/retry: feature flag new retry behavior 2024-05-24 20:24:02 +02:00
Michael Eischer b1266867d2 repository: wait max 1 minutes for lock removal if context is canceled
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.
2024-05-24 20:24:02 +02:00
Michael Eischer 98709a4372 retry: reduce total number of retries
Retries in restic try to solve two main problems:
- retry a temporarily failed operation
- tolerate temporary network interruptions

The first problem only requires a few retries, whereas the last one benefits
primarily from spreading the requests over a longer duration.

Increasing the default multiplier and the initial interval works for
both cases. The first few retries only take a few seconds, while later
retries quickly reach the maximum interval of one minute. This ensures
that the total number of retries issued by restic will remain at around
21 retries for a 15 minute period. As the concurrency in restic is
bounded, retries drastically reduce the number of requests sent to a
backend. This helps to prevent overloading the backend.
2024-05-24 20:24:02 +02:00
Michael Eischer 512cd6ef07 retry: ensure that there's always at least one retry
Previously, if an operation failed after 15 minutes, then it would never
be retried. This means that large backend requests are more unreliable
than smaller ones.
2024-05-24 20:24:02 +02:00
Michael Eischer a60ee9b764 retry: limit retries based on elapsed time not count
Depending on how long an operation takes to fail, the total retry
duration can currently vary between 1.5 and 15 minutes. In particular
for temporarily interrupted network connections, the former timeout is
too short. Thus always use a limit of 15 minutes.
2024-05-24 20:24:02 +02:00
Michael Eischer a3633cad9e retry: explicitly log failed requests
This simplifies finding the request in the log output that cause an
operation to fail.
2024-05-24 20:24:02 +02:00
greatroar 66d03c797e ui, termstatus: Move WrapStdio
Saves some imports. ui still needs to import ui/termstatus from
message.go.
2024-05-21 11:25:16 +02:00
greatroar 0b56214473 ui: Simplify stdio wrapper
The StdioWrapper type is really just a pair of io.WriteClosers, so
remove it in favor of a function that returns two of those. Test
coverage increases because the removed code was not tested.
2024-05-21 11:23:32 +02:00
greatroar 7f439a9c34 ui: Inline lineWriter into StdioWrapper 2024-05-21 11:00:09 +02:00
Michael Eischer c56ecec9aa azure: deduplicate cli and default credentials case 2024-05-18 22:15:54 +02:00
Maik Riechert 355f520936 Azure: add option to force use of CLI credential 2024-05-18 22:15:54 +02:00
Michael Eischer 223aa22cb0 replace some uses of restic.Repository with finegrained interfaces 2024-05-18 21:42:51 +02:00
Michael Eischer 291c9677de restic/repository: remove Backend() method 2024-05-18 21:42:51 +02:00
Michael Eischer 673496b091 repository: clean cache between CheckPack retries
The cache cleanup pattern is also used in ListPack etc.
2024-05-18 21:42:51 +02:00
Michael Eischer 3d2410ed50 Replace some repo.RemoveUnpacked usages
These will eventually be blocked as they do not delete Snapshots.
2024-05-18 21:42:51 +02:00
Michael Eischer d2c26e33f3 repository: remove further usages of repo.Backend() 2024-05-18 21:42:51 +02:00
Michael Eischer 8a425c2f0a remove usages of repo.Backend() from tests 2024-05-18 21:42:51 +02:00
Michael Eischer aa4647f773 repository: unexport PackBlobIterator 2024-05-18 21:42:51 +02:00
Michael Eischer 94e863885c check: move verification of individual pack file to repository 2024-05-18 21:42:50 +02:00
Michael Eischer e40943a75d restic: remove backend usage from lock test 2024-05-18 21:38:31 +02:00
Michael Eischer 67e2ba0d40 repository: Lock requires *repository.Repository
This allows the Lock function to access the backend, even once the
Backend method is removed from the interface.
2024-05-18 21:38:31 +02:00
Michael Eischer d8b184b3d3 repository: convert test helper to return *repository.Repository 2024-05-18 21:38:31 +02:00
Michael Eischer a1ca5e15c4 migrations: add temporary hack for s3_layout
The migration will be removed after the next restic release anyways.
Thus, there's no need for a clean implementation.
2024-05-18 21:38:31 +02:00
Michael Eischer 34d90aecf9 migrations: move logic of upgrade_repo_v2 to repository package
The migration modifies repository internals and thus should live within
the repository package.
2024-05-18 21:38:31 +02:00
Michael Eischer ab9077bc13 replace usages of backend.Remove() with repository.RemoveUnpacked()
RemoveUnpacked will eventually block removal of all filetypes other than
snapshots. However, getting there requires a major refactor to provide
some components with privileged access.
2024-05-18 21:38:31 +02:00
Michael Eischer 0c1ba6d95d backend: remove unused Location method 2024-05-18 21:38:31 +02:00
Michael Eischer 74d90653e0 check: use ReadFull to load pack header in checkPack
This ensures that the pack header is actually read completely.
Previously, for a truncated file it was possible to only read a part of
the header, as backend.Load(...) is not guaranteed to return as many
bytes as requested by the length parameter.
2024-05-18 21:28:54 +02:00
Michael Eischer 8f8d872a68 fix compatibility with go 1.19 2024-05-18 21:28:54 +02:00
Michael Eischer ff0744b3af check: test checkPack retries 2024-05-18 21:28:54 +02:00
Michael Eischer 987c3b250c repository: test retries of ListPack 2024-05-18 21:28:54 +02:00
Michael Eischer bf16096771 repository: test LoadBlob retries 2024-05-18 21:28:54 +02:00
Michael Eischer 4f45668b7c repository: rework and extend LoadRaw tests 2024-05-18 21:28:54 +02:00
Michael Eischer ac805d6838 cache: cleanup debug logs 2024-05-18 21:28:54 +02:00
Michael Eischer 5214af88e2 cache: test forget behavior 2024-05-18 21:28:54 +02:00
Michael Eischer 3ff063e913 check: verify pack a second time if broken 2024-05-18 21:28:54 +02:00
Michael Eischer 385cee09dc repository: fix caching of tree packs in LoadBlobsFromPack 2024-05-18 21:28:54 +02:00
Michael Eischer e734746f75 cache: forget cached file at most once
This is inspired by the circuit breaker pattern used for distributed
systems. If too many requests fails, then it is better to immediately
fail new requests for a limited time to give the backend time to
recover.

By only forgetting a file in the cache at most once, we can ensure that
a broken file is only retrieved once again from the backend. If the file
stored there is broken, previously it would be cached and deleted
continuously. Now, it is retrieved only once again, all later requests
just use the cached copy and either succeed or fail immediately.
2024-05-18 21:28:54 +02:00
Michael Eischer 97a307df1a cache: Always use cached file if it exists
A file is always cached whole. Thus, any out of bounds access will also
fail when directed at the backend. To handle case in which the cached
file is broken, then caller must call Cache.Forget(h) for the file in
question.
2024-05-18 21:28:54 +02:00
Michael Eischer 433a6aad29 repository: remove redundant blob loading fallback from RepairPacks
LoadBlobsFromPack already implements the same fallback behavior.
2024-05-18 21:28:54 +02:00
Michael Eischer e401af07b2 check: fix error message formatting 2024-05-18 21:28:54 +02:00
Michael Eischer 7017adb7e9 repository: retry failed ListPack once 2024-05-18 21:28:54 +02:00
Michael Eischer e33ce7f408 repository: retry failed LoadBlob once 2024-05-18 21:28:54 +02:00
Michael Eischer 2ace242f36 repository: make reloading broken files explicit 2024-05-18 21:28:54 +02:00
Michael Eischer e9390352a7 cache: code cleanups 2024-05-18 21:26:00 +02:00
Michael Eischer 503c8140b1 repository: unify blob decoding code 2024-05-18 21:26:00 +02:00
Michael Eischer 6563f1d2ca repository: remove redundant debug log 2024-05-18 21:26:00 +02:00
Michael Eischer 021fb49559 repository: Implement repository.LoadUnpacked using LoadRaw
Both functions were using a similar implementation.
2024-05-18 21:26:00 +02:00
Michael Eischer 779c8d3527 debug/repair packs/upgrade repo v2: use repository.LoadRaw
This replaces calling the low-level backend.Load() method.
2024-05-18 21:26:00 +02:00
Michael Eischer 1d6d3656b0 repository: move backend.LoadAll to repository.LoadRaw
LoadRaw also includes improved context cancellation handling similar to the
implementation in repository.LoadUnpacked.

The removed cache backend test will be added again later on.
2024-05-18 21:26:00 +02:00
Michael Eischer 47232bf8b0 backend: move LimitReadCloser to util package
The helper is only intended for usage by backend implementations.
2024-05-18 21:26:00 +02:00
Michael Eischer 53d15bcd1b retry: add circuit breaker to load method
If a file exhausts its retry attempts, then it is likely not accessible
the next time. Thus, immediately fail all load calls for that file to
avoid useless retries.
2024-05-18 19:59:26 +02:00
Michael Eischer 394c8ca3ed rest/rclone/s3/sftp/swift: move short file detection behind feature gate
These backends tend to use a large variety of server implementations.
Some of those implementations might prove problematic with the new
checks.
2024-05-18 19:59:26 +02:00
Michael Eischer 6328b7e1f5 replace "too small" with "too short" in error messages 2024-05-18 19:59:26 +02:00
Michael Eischer aeb7eb245c retry: do not retry permanent errors
This is currently gated behind a feature flag as some unexpected
interactions might show up in the wild.
2024-05-18 19:59:26 +02:00
Michael Eischer bf8cc59889 Use generic backend-error-redesign feature flag instead of http-timeouts
An individual flag for each change of the backend error handling would
be too finegrained. Thus, add a generic flag.
2024-05-18 19:54:52 +02:00
Michael Eischer 4740528a0b backend: add tests for IsPermanentError 2024-05-18 19:54:52 +02:00
Michael Eischer 6a85df7297 backend: add IsPermanentError() method to interface 2024-05-18 19:54:52 +02:00
Michael Eischer cfc420664a mem: stricter handling of out of bounds requests 2024-05-18 19:54:52 +02:00
Michael Eischer d40f23e716 azure/b2/gs/s3/swift: adapt cloud backend 2024-05-18 19:54:51 +02:00
Michael Eischer e793c002ec local: stricter handling of short files 2024-05-18 19:54:21 +02:00
Michael Eischer b4895ebd76 rest: rework error reporting and report too short files 2024-05-18 19:54:21 +02:00
Michael Eischer eaa3f81d6b sftp: check for truncated files without an extra backend request 2024-05-18 19:54:21 +02:00
Michael Eischer c6d74458ee sftp: improve handling of too short files 2024-05-18 19:54:21 +02:00
aneesh-n 9d1b23964d
Update comment for licensing 2024-05-17 14:19:05 -06:00