Before this change, resuming from an unencrypted dataset with
send.raw=true specified wouldn't work with zrepl due to overly
restrictive resume token checking.
An initial PR to fix this was made in https://github.com/zrepl/zrepl/pull/503
but it didn't address the core of the problem.
The core of the problem was that zrepl assumed that if a resume token
contained `rawok=true, compressok=true`, the resulting send would be
encrypted. But if the sender dataset was unencrypted, such a resume would
actually result in an unencrypted send.
Which could be totally legitimate but zrepl failed to recognize that.
BACKGROUND
==========
The following snippets of OpenZFS code are insightful regarding how the
various ${X}ok values in the resume token are handled:
- 6c3c5fcfbe/module/zfs/dmu_send.c (L1947-L2012)
- 6c3c5fcfbe/module/zfs/dmu_recv.c (L877-L891)
- https://github.com/openzfs/zfs/blob/6c3c5fc/lib/libzfs/libzfs_sendrecv.c#L1663-L1672
Basically, some zfs send flags make the DMU send code set some DMU send
stream featureflags, although it's not a pure mapping, i.e, which DMU
send stream flags are used depends somewhat on the dataset (e.g., is it
encrypted or not, or, does it use zstd or not).
Then, the receiver looks at some (but not all) feature flags and maps
them to ${X}ok dataset zap attributes.
These are funnelled back to the sender 1:1 through the resume_token.
And the sender turns them into lzc flags.
As an example, let's look at zfs send --raw.
if the sender requests a raw send on an unencrypted dataset, the send
stream (and hence the resume token) will not have the raw stream
featureflag set, and hence the resume token will not have the rawok
field set. Instead, it will have compressok, embedok, and depending
on whether large blocks are present in the dataset, largeblockok set.
WHAT'S ZREPL'S ROLE IN THIS?
============================
zrepl provides a virtual encrypted sendflag that is like `raw`,
but further ensures that we only send encrypted datasets.
For any other resume token stuff, it shoudn't do any checking,
because it's a futile effort to keep up with ZFS send/recv features
that are orthogonal to encryption.
CHANGES MADE IN THIS COMMIT
===========================
- Rip out a bunch of needless checking that zrepl would do during
planning. These checks were there to give better error messages,
but actually, the error messages created by the endpoint.Sender.Send
RPC upon send args validation failure are good enough.
- Add platformtests to validate all combinations of
(Unencrypted/Encrypted FS) x (send.encrypted = true | false) x (send.raw = true | false)
for cases both non-resuming and resuming send.
Additional manual testing done:
1. With zrepl 0.5, setup with unencrypted dataset, send.raw=true specified, no send.encrypted specified.
2. Observe that regular non-resuming send works, but resuming doesn't work.
3. Upgrade zrepl to this change.
4. Observe that both regular and resuming send works.
closes https://github.com/zrepl/zrepl/pull/613
fixes https://github.com/zrepl/zrepl/issues/504
Problem:
plain send + recv with root_fs encrypted + placeholders causes plain recvs
whereas user would expect encrypt-on-recv
Reason:
We create placeholder filesytems with -o encryption=off.
Thus, children received below those placeholders won't inherit
encryption of root_fs.
Fix:
We'll have three values for `recv.placeholders.encryption: unspecified (default) | off | inherit`.
When we create a placeholder, we will fail the operation if `recv.placeholders.encryption = unspecified`.
The exception is if the placeholder filesystem is to encode the client identity ($root_fs/$client_identity) in a pull job.
Those are created in `inherit` mode if the config field is `unspecified` so that users who don't need
placeholders are not bothered by these details.
Future Work:
Automatically warn existing users of encrypt-on-recv about the problem
if they are affected.
The problem that I hit during implementation of this is that the
`encryption` prop's `source` doesn't quite behave like other props:
`source` is `default` for `encryption=off` and `-` when `encryption=on`.
Hence, we can't use `source` to distinguish the following 2x2 cases:
(1) placeholder created with explicit -o encryption=off
(2) placeholder created without specifying -o encryption
with
(A) an encrypted parent at creation time
(B) an unencrypted parent at creation time
This commit was motivated by https://github.com/zrepl/zrepl/issues/495
where, on FreeBSD with OpenZFS 2.0, a SendStream.Close() call might wait indefinitely for `zfs send` to exit.
The reason is that, due to the refactoring done for redacted send & recv
(30af21b025),
the `dump_bytes` function, which writes to the pipe, executes in a separate thread (synctask taskq) iff not `HAVE_LARGE_STACKS`.
The `zfs send` process/thread waits for that taskq thread using an uninterruptible primitive.
So when we SIGKILL `zfs send`, that signal doesn't reach the right thread to interrupt the pipe write.
Theoretically this affects both Linux and FreeBSD, but most Linux users `HAVE_LARGE_STACKS` and since https://github.com/penzfs/zfs/pull/12350/files OpenZFS on FreeBSD `HAVE_LARGE_STACKS` as well.
However, at least until FreeBSD 13.1, possibly for the entire 13 lifecycle, we're going to have to live with that oddity.
Measures taken in this commit:
- Report the behavior as an upstream bug https://github.com/openzfs/zfs/issues/12500
- Change SendStream code so that it closes zrepl's read-end of the pipe (see comment in code)
- Clean up and make explicit SendStream's state handling
- Write extensive platformtests for SendStream
- They pass on my Linux install and on FreeBSD 12
- FreeBSD 13 still needs testing.
fixes https://github.com/zrepl/zrepl/issues/495
Addition by @problame: move getPipeCapacityHint() into platform-specific
code. This has the added benefit of not recognizing the envvar as an
envconst on platform that do not support resizing pipes. => won't show
up in (zrepl status --raw).Global.Envconst
fixes#424
cloes #449
This commit
- adds a configuration in which no step holds, replication cursors, etc. are created
- removes the send.step_holds.disable_incremental setting
- creates a new config option `replication` for active-side jobs
- adds the replication.protection.{initial,incremental} settings, each
of which can have values
- `guarantee_resumability`
- `guarantee_incremental`
- `guarantee_nothing`
(refer to docs/configuration/replication.rst for semantics)
The `replication` config from an active side is sent to both endpoint.Sender and endpoint.Receiver
for each replication step. Sender and Receiver then act accordingly.
For `guarantee_incremental`, we add the new `tentative-replication-cursor` abstraction.
The necessity for that abstraction is outlined in https://github.com/zrepl/zrepl/issues/340.
fixes https://github.com/zrepl/zrepl/issues/340
- drop HintMostRecentCommonAncestor rpc call
- it is wrong to put faith into the active side of the replication to always make that call
(we might not trust it, ref pull setup)
- clean up step holds + step bookmarks + replication cursor bookmarks on
send RPC instead
- this makes it symmetric with Receive RPC
- use a cache (endpoint.sendAbstractionsCache) to avoid the cost of
listing the on-disk endpoint abstractions state on every step
The "create" methods for endpoint abstractions (CreateReplicationCursor, HoldStep) are now fully
idempotent and return an Abstraction.
Notes about endpoint.sendAbstractionsCache:
- fills lazily from disk state on first `Get` operation
- fill from disk is generally only attempted once
- unless the `ListAbstractions` fails, in which case the fill from
disk is retried on next `Get` (the current `Get` will observe a
subset of the actual on-disk abstractions)
- the `Invalidate` method is called
- it is a global (zrepl process-wide) cache
fixes#316
The motivation for this recatoring are based on two independent issues:
- @JMoVS found that the changes merged as part of #259 slowed his OS X
based installation down significantly.
Analysis of the zfs command logging introduced in #296 showed that
`zfs holds` took most of the execution time, and they pointed out
that not all of those `zfs holds` invocations were actually necessary.
I.e.: zrepl was inefficient about retrieving information from ZFS.
- @InsanePrawn found that failures on initial replication would lead
to step holds accumulating on the sending side, i.e. they would never
be cleaned up in the HintMostRecentCommonAncestor RPC handler.
That was because we only sent that RPC if there was a most recent
common ancestor detected during replication planning.
@InsanePrawn prototyped an implementation of a `zrepl zfs-abstractions release`
command to mitigate the situation.
As part of that development work and back-and-forth with @problame,
it became evident that the abstractions that #259 built on top of
zfs in package endpoint (step holds, replication cursor,
last-received-hold), were not well-represented for re-use in the
`zrepl zfs-abstractions release` subocommand prototype.
This commit refactors package endpoint to address both of these issues:
- endpoint abstractions now share an interface `Abstraction` that, among
other things, provides a uniform `Destroy()` method.
However, that method should not be destroyed directly but instead
the package-level `BatchDestroy` function should be used in order
to allow for a migration to zfs channel programs in the future.
- endpoint now has a query facitilty (`ListAbstractions`) which is
used to find on-disk
- step holds and bookmarks
- replication cursors (v1, v2)
- last-received-holds
By describing the query in a struct, we can centralized the retrieval
of information via the ZFS CLI and only have to be clever once.
We are "clever" in the following ways:
- When asking for hold-based abstractions, we only run `zfs holds` on
snapshot that have `userrefs` > 0
- To support this functionality, add field `UserRefs` to zfs.FilesystemVersion
and retrieve it anywhere we retrieve zfs.FilesystemVersion from ZFS.
- When asking only for bookmark-based abstractions, we only run
`zfs list -t bookmark`, not with snapshots.
- Currently unused (except for CLI) per-filesystem concurrent lookup
- Option to only include abstractions with CreateTXG in a specified range
- refactor `endpoint`'s various ZFS info retrieval methods to use
`ListAbstractions`
- rename the `zrepl holds list` command to `zrepl zfs-abstractions list`
- make `zrepl zfs-abstractions list` consume endpoint.ListAbstractions
- Add a `ListStale` method which, given a query template,
lists stale holds and bookmarks.
- it uses replication cursor has different modes
- the new `zrepl zfs-abstractions release-{all,stale}` commands can be used
to remove abstractions of package endpoint
- Adjust HintMostRecentCommonAncestor RPC for stale-holds cleanup:
- send it also if no most recent common ancestor exists between sender and receiver
- have the sender clean up its abstractions when it receives the RPC
with no most recent common ancestor, using `ListStale`
- Due to changed semantics, bump the protocol version.
- Adjust HintMostRecentCommonAncestor RPC for performance problems
encountered by @JMoVS
- by default, per (job,fs)-combination, only consider cleaning
step holds in the createtxg range
`[last replication cursor,conservatively-estimated-receive-side-version)`
- this behavior ensures resumability at cost proportional to the
time that replication was donw
- however, as explained in a comment, we might leak holds if
the zrepl daemon stops running
- that trade-off is acceptable because in the presumably rare
this might happen the user has two tools at their hand:
- Tool 1: run `zrepl zfs-abstractions release-stale`
- Tool 2: use env var `ZREPL_ENDPOINT_SENDER_HINT_MOST_RECENT_STEP_HOLD_CLEANUP_MODE`
to adjust the lower bound of the createtxg range (search for it in the code).
The env var can also be used to disable hold-cleanup on the
send-side entirely.
supersedes closes#293
supersedes closes#282fixes#280fixes#278
Additionaly, we fixed a couple of bugs:
- zfs: fix half-nil error reporting of dataset-does-not-exist for ZFSListChan and ZFSBookmark
- endpoint: Sender's `HintMostRecentCommonAncestor` handler would not
check whether access to the specified filesystem was allowed.
- **Resumable Send & Recv Support**
No knobs required, automatically used where supported.
- **Hold-Protected Send & Recv**
Automatic ZFS holds to ensure that we can always resume a replication step.
- **Encrypted Send & Recv Support** for OpenZFS native encryption.
Configurable at the job level, i.e., for all filesystems a job is responsible for.
- **Receive-side hold on last received dataset**
The counterpart to the replication cursor bookmark on the send-side.
Ensures that incremental replication will always be possible between a sender and receiver.
Design Doc
----------
`replication/design.md` doc describes how we use ZFS holds and bookmarks to ensure that a single replication step is always resumable.
The replication algorithm described in the design doc introduces the notion of job IDs (please read the details on this design doc).
We reuse the job names for job IDs and use `JobID` type to ensure that a job name can be embedded into hold tags, bookmark names, etc.
This might BREAK CONFIG on upgrade.
Protocol Version Bump
---------------------
This commit makes backwards-incompatible changes to the replication/pdu protobufs.
Thus, bump the version number used in the protocol handshake.
Replication Cursor Format Change
--------------------------------
The new replication cursor bookmark format is: `#zrepl_CURSOR_G_${this.GUID}_J_${jobid}`
Including the GUID enables transaction-safe moving-forward of the cursor.
Including the job id enables that multiple sending jobs can send the same filesystem without interfering.
The `zrepl migrate replication-cursor:v1-v2` subcommand can be used to safely destroy old-format cursors once zrepl has created new-format cursors.
Changes in This Commit
----------------------
- package zfs
- infrastructure for holds
- infrastructure for resume token decoding
- implement a variant of OpenZFS's `entity_namecheck` and use it for validation in new code
- ZFSSendArgs to specify a ZFS send operation
- validation code protects against malicious resume tokens by checking that the token encodes the same send parameters that the send-side would use if no resume token were available (i.e. same filesystem, `fromguid`, `toguid`)
- RecvOptions support for `recv -s` flag
- convert a bunch of ZFS operations to be idempotent
- achieved through more differentiated error message scraping / additional pre-/post-checks
- package replication/pdu
- add field for encryption to send request messages
- add fields for resume handling to send & recv request messages
- receive requests now contain `FilesystemVersion To` in addition to the filesystem into which the stream should be `recv`d into
- can use `zfs recv $root_fs/$client_id/path/to/dataset@${To.Name}`, which enables additional validation after recv (i.e. whether `To.Guid` matched what we received in the stream)
- used to set `last-received-hold`
- package replication/logic
- introduce `PlannerPolicy` struct, currently only used to configure whether encrypted sends should be requested from the sender
- integrate encryption and resume token support into `Step` struct
- package endpoint
- move the concepts that endpoint builds on top of ZFS to a single file `endpoint/endpoint_zfs.go`
- step-holds + step-bookmarks
- last-received-hold
- new replication cursor + old replication cursor compat code
- adjust `endpoint/endpoint.go` handlers for
- encryption
- resumability
- new replication cursor
- last-received-hold
- client subcommand `zrepl holds list`: list all holds and hold-like bookmarks that zrepl thinks belong to it
- client subcommand `zrepl migrate replication-cursor:v1-v2`
We assumed that `zfs recv -F FS` would basically replace FS inplace, leaving its children untouched.
That is in fact not the case, it only works if `zfs send -R` is set, which we don't do.
Thus, implement the required functionality manually.
This solves a `zfs recv` error that would occur when a filesystem previously created as placeholder on the receiving side becomes a non-placeholder filesystem (likely due to config change on the sending side):
zfs send pool1/foo@1 | zfs recv -F pool1/bar
cannot receive new filesystem stream:
destination has snapshots (eg. pool1/bar)
must destroy them to overwrite it
A bookmark with a well-known name is used to track which version was
last successfully received by the receiver.
The createtxg that can be retrieved from the bookmark using `zfs get` is
used to set the Replicated attribute of each snap on the sender:
If the snap's CreateTXG > the cursor's, it is not yet replicated,
otherwise it has been.
There is an optional config option to change the behvior to
`CreateTXG >= the cursor's`, and the implementation defaults to that.
The reason: While things work just fine with `CreateTXG > the cursor's`,
ZFS does not provide size estimates in a `zfs send` dry run
(see acd2418).
However, to enable the use case of keeping the snapshot only around for
the replication, the config flag exists.
Since we don't implement screen-scraping of ZFS output ATM, this is
better than nothing, as user's may be able to figure out what' sthe
problem from the logs / status reports.
