2018-05-02 21:26:56 +02:00
|
|
|
package zfs
|
|
|
|
|
|
|
|
import (
|
|
|
|
"context"
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
"fmt"
|
2018-05-02 21:26:56 +02:00
|
|
|
"os/exec"
|
|
|
|
"regexp"
|
|
|
|
"strconv"
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
"strings"
|
|
|
|
"sync"
|
2018-05-02 21:26:56 +02:00
|
|
|
"time"
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
|
|
|
|
"github.com/pkg/errors"
|
|
|
|
|
|
|
|
"github.com/zrepl/zrepl/util/envconst"
|
2020-03-27 12:35:57 +01:00
|
|
|
"github.com/zrepl/zrepl/zfs/zfscmd"
|
2018-05-02 21:26:56 +02:00
|
|
|
)
|
|
|
|
|
2020-02-23 23:24:12 +01:00
|
|
|
// NOTE: Update ZFSSendARgs.Validate when changing fields (potentially SECURITY SENSITIVE)
|
2018-05-02 21:26:56 +02:00
|
|
|
type ResumeToken struct {
|
2020-11-07 17:52:10 +01:00
|
|
|
HasFromGUID, HasToGUID bool
|
|
|
|
FromGUID, ToGUID uint64
|
|
|
|
ToName string
|
|
|
|
HasCompressOK, CompressOK bool
|
|
|
|
HasRawOk, RawOK bool
|
|
|
|
HasLargeBlockOK, LargeBlockOK bool
|
|
|
|
HasEmbedOk, EmbedOK bool
|
|
|
|
HasSavedOk, SavedOk bool
|
2018-05-02 21:26:56 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
var resumeTokenNVListRE = regexp.MustCompile(`\t(\S+) = (.*)`)
|
2020-02-23 23:24:12 +01:00
|
|
|
|
2018-05-02 21:26:56 +02:00
|
|
|
var resumeTokenContentsRE = regexp.MustCompile(`resume token contents:\nnvlist version: 0`)
|
|
|
|
var resumeTokenIsCorruptRE = regexp.MustCompile(`resume token is corrupt`)
|
|
|
|
|
|
|
|
var ResumeTokenCorruptError = errors.New("resume token is corrupt")
|
|
|
|
var ResumeTokenDecodingNotSupported = errors.New("zfs binary does not allow decoding resume token or zrepl cannot scrape zfs output")
|
|
|
|
var ResumeTokenParsingError = errors.New("zrepl cannot parse resume token values")
|
|
|
|
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
var resumeSendSupportedCheck struct {
|
|
|
|
once sync.Once
|
|
|
|
supported bool
|
|
|
|
err error
|
|
|
|
}
|
|
|
|
|
2020-03-27 00:57:33 +01:00
|
|
|
func ResumeSendSupported(ctx context.Context) (bool, error) {
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
resumeSendSupportedCheck.once.Do(func() {
|
|
|
|
// "feature discovery"
|
2020-03-27 12:35:57 +01:00
|
|
|
cmd := zfscmd.CommandContext(ctx, "zfs", "send")
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
output, err := cmd.CombinedOutput()
|
|
|
|
if ee, ok := err.(*exec.ExitError); !ok || ok && !ee.Exited() {
|
|
|
|
resumeSendSupportedCheck.err = errors.Wrap(err, "resumable send cli support feature check failed")
|
|
|
|
}
|
|
|
|
def := strings.Contains(string(output), "receive_resume_token")
|
|
|
|
resumeSendSupportedCheck.supported = envconst.Bool("ZREPL_EXPERIMENTAL_ZFS_SEND_RESUME_SUPPORTED", def)
|
|
|
|
debug("resume send feature check complete %#v", &resumeSendSupportedCheck)
|
|
|
|
})
|
|
|
|
return resumeSendSupportedCheck.supported, resumeSendSupportedCheck.err
|
|
|
|
}
|
|
|
|
|
|
|
|
var resumeRecvPoolSupportRecheckTimeout = envconst.Duration("ZREPL_ZFS_RESUME_RECV_POOL_SUPPORT_RECHECK_TIMEOUT", 30*time.Second)
|
|
|
|
|
|
|
|
type resumeRecvPoolSupportedResult struct {
|
|
|
|
lastCheck time.Time
|
|
|
|
supported bool
|
|
|
|
err error
|
|
|
|
}
|
|
|
|
|
|
|
|
var resumeRecvSupportedCheck struct {
|
|
|
|
mtx sync.RWMutex
|
|
|
|
flagSupport struct {
|
|
|
|
checked bool
|
|
|
|
supported bool
|
|
|
|
err error
|
|
|
|
}
|
|
|
|
poolSupported map[string]resumeRecvPoolSupportedResult
|
|
|
|
}
|
|
|
|
|
|
|
|
// fs == nil only checks for CLI support
|
|
|
|
func ResumeRecvSupported(ctx context.Context, fs *DatasetPath) (bool, error) {
|
|
|
|
sup := &resumeRecvSupportedCheck
|
|
|
|
sup.mtx.RLock()
|
|
|
|
defer sup.mtx.RUnlock()
|
|
|
|
upgradeWhile := func(cb func()) {
|
|
|
|
sup.mtx.RUnlock()
|
|
|
|
defer sup.mtx.RLock()
|
|
|
|
sup.mtx.Lock()
|
|
|
|
defer sup.mtx.Unlock()
|
|
|
|
cb()
|
|
|
|
}
|
|
|
|
|
|
|
|
if !sup.flagSupport.checked {
|
2020-03-27 12:35:57 +01:00
|
|
|
output, err := zfscmd.CommandContext(ctx, ZFS_BINARY, "receive").CombinedOutput()
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
upgradeWhile(func() {
|
|
|
|
sup.flagSupport.checked = true
|
|
|
|
if ee, ok := err.(*exec.ExitError); err != nil && (!ok || ok && !ee.Exited()) {
|
|
|
|
sup.flagSupport.err = err
|
|
|
|
} else {
|
|
|
|
sup.flagSupport.supported = strings.Contains(string(output), "-A <filesystem|volume>")
|
|
|
|
}
|
|
|
|
debug("resume recv cli flag feature check result: %#v", sup.flagSupport)
|
|
|
|
})
|
|
|
|
// fallthrough
|
|
|
|
}
|
|
|
|
|
|
|
|
if sup.flagSupport.err != nil {
|
|
|
|
return false, errors.Wrap(sup.flagSupport.err, "zfs recv feature check for resumable send & recv failed")
|
|
|
|
} else if !sup.flagSupport.supported || fs == nil {
|
|
|
|
return sup.flagSupport.supported, nil
|
|
|
|
}
|
|
|
|
|
|
|
|
// Flag is supported and pool-support is request
|
|
|
|
// Now check for pool support
|
|
|
|
|
|
|
|
pool, err := fs.Pool()
|
|
|
|
if err != nil {
|
|
|
|
return false, errors.Wrap(err, "resume recv check requires pool of dataset")
|
|
|
|
}
|
|
|
|
|
|
|
|
if sup.poolSupported == nil {
|
|
|
|
upgradeWhile(func() {
|
|
|
|
sup.poolSupported = make(map[string]resumeRecvPoolSupportedResult)
|
|
|
|
})
|
|
|
|
}
|
|
|
|
|
|
|
|
var poolSup resumeRecvPoolSupportedResult
|
|
|
|
var ok bool
|
|
|
|
if poolSup, ok = sup.poolSupported[pool]; !ok || // shadow
|
|
|
|
(!poolSup.supported && time.Since(poolSup.lastCheck) > resumeRecvPoolSupportRecheckTimeout) {
|
|
|
|
|
2020-03-27 12:35:57 +01:00
|
|
|
output, err := zfscmd.CommandContext(ctx, "zpool", "get", "-H", "-p", "-o", "value", "feature@extensible_dataset", pool).CombinedOutput()
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
if err != nil {
|
|
|
|
debug("resume recv pool support check result: %#v", sup.flagSupport)
|
|
|
|
poolSup.supported = false
|
|
|
|
poolSup.err = err
|
|
|
|
} else {
|
|
|
|
poolSup.err = nil
|
|
|
|
o := strings.TrimSpace(string(output))
|
|
|
|
poolSup.supported = o == "active" || o == "enabled"
|
|
|
|
}
|
|
|
|
poolSup.lastCheck = time.Now()
|
|
|
|
|
|
|
|
// we take the lock late, so two updaters might check simultaneously, but that shouldn't hurt
|
|
|
|
upgradeWhile(func() {
|
|
|
|
sup.poolSupported[pool] = poolSup
|
|
|
|
})
|
|
|
|
// fallthrough
|
|
|
|
}
|
|
|
|
|
|
|
|
if poolSup.err != nil {
|
|
|
|
return false, errors.Wrapf(poolSup.err, "pool %q check for feature@extensible_dataset feature failed", pool)
|
|
|
|
}
|
|
|
|
|
|
|
|
return poolSup.supported, nil
|
|
|
|
}
|
|
|
|
|
2018-05-02 21:26:56 +02:00
|
|
|
// Abuse 'zfs send' to decode the resume token
|
|
|
|
//
|
|
|
|
// FIXME: implement nvlist unpacking in Go and read through libzfs_sendrecv.c
|
|
|
|
func ParseResumeToken(ctx context.Context, token string) (*ResumeToken, error) {
|
|
|
|
|
2020-03-27 00:57:33 +01:00
|
|
|
if supported, err := ResumeSendSupported(ctx); err != nil {
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
return nil, err
|
|
|
|
} else if !supported {
|
|
|
|
return nil, ResumeTokenDecodingNotSupported
|
|
|
|
}
|
|
|
|
|
2018-05-02 21:26:56 +02:00
|
|
|
// Example resume tokens:
|
|
|
|
//
|
|
|
|
// From a non-incremental send
|
|
|
|
// 1-bf31b879a-b8-789c636064000310a500c4ec50360710e72765a5269740f80cd8e4d3d28a534b18e00024cf86249f5459925acc802a8facbf243fbd3433858161f5ddb9ab1ae7c7466a20c97382e5f312735319180af2f3730cf58166953824c2cc0200cde81651
|
|
|
|
|
|
|
|
// From an incremental send
|
|
|
|
// 1-c49b979a2-e0-789c636064000310a501c49c50360710a715e5e7a69766a63040c1eabb735735ce8f8d5400b2d991d4e52765a5269740f82080219f96569c5ac2000720793624f9a4ca92d46206547964fd25f91057f09e37babb88c9bf5503499e132c9f97989bcac050909f9f63a80f34abc421096616007c881d4c
|
|
|
|
|
|
|
|
// Resulting output of zfs send -nvt <token>
|
|
|
|
//
|
|
|
|
//resume token contents:
|
|
|
|
//nvlist version: 0
|
|
|
|
// fromguid = 0x595d9f81aa9dddab
|
|
|
|
// object = 0x1
|
|
|
|
// offset = 0x0
|
|
|
|
// bytes = 0x0
|
|
|
|
// toguid = 0x854f02a2dd32cf0d
|
|
|
|
// toname = pool1/test@b
|
|
|
|
//cannot resume send: 'pool1/test@b' used in the initial send no longer exists
|
|
|
|
|
2020-03-27 12:35:57 +01:00
|
|
|
cmd := zfscmd.CommandContext(ctx, ZFS_BINARY, "send", "-nvt", string(token))
|
2018-05-02 21:26:56 +02:00
|
|
|
output, err := cmd.CombinedOutput()
|
|
|
|
if err != nil {
|
|
|
|
if exitErr, ok := err.(*exec.ExitError); ok {
|
|
|
|
if !exitErr.Exited() {
|
|
|
|
return nil, err
|
|
|
|
}
|
|
|
|
// we abuse zfs send for decoding, the exit error may be due to
|
|
|
|
// a) the token being from a third machine
|
|
|
|
// b) it no longer exists on the machine where
|
|
|
|
} else {
|
|
|
|
return nil, err
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if !resumeTokenContentsRE.Match(output) {
|
|
|
|
if resumeTokenIsCorruptRE.Match(output) {
|
|
|
|
return nil, ResumeTokenCorruptError
|
|
|
|
}
|
|
|
|
return nil, ResumeTokenDecodingNotSupported
|
|
|
|
}
|
|
|
|
|
|
|
|
matches := resumeTokenNVListRE.FindAllStringSubmatch(string(output), -1)
|
|
|
|
if matches == nil {
|
|
|
|
return nil, ResumeTokenDecodingNotSupported
|
|
|
|
}
|
|
|
|
|
|
|
|
rt := &ResumeToken{}
|
|
|
|
|
|
|
|
for _, m := range matches {
|
|
|
|
attr, val := m[1], m[2]
|
|
|
|
switch attr {
|
|
|
|
case "fromguid":
|
|
|
|
rt.FromGUID, err = strconv.ParseUint(val, 0, 64)
|
|
|
|
if err != nil {
|
|
|
|
return nil, ResumeTokenParsingError
|
|
|
|
}
|
|
|
|
rt.HasFromGUID = true
|
|
|
|
case "toguid":
|
|
|
|
rt.ToGUID, err = strconv.ParseUint(val, 0, 64)
|
|
|
|
if err != nil {
|
|
|
|
return nil, ResumeTokenParsingError
|
|
|
|
}
|
|
|
|
rt.HasToGUID = true
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
case "toname":
|
|
|
|
rt.ToName = val
|
|
|
|
case "rawok":
|
|
|
|
rt.HasRawOk = true
|
|
|
|
rt.RawOK, err = strconv.ParseBool(val)
|
|
|
|
if err != nil {
|
|
|
|
return nil, ResumeTokenParsingError
|
|
|
|
}
|
|
|
|
case "compressok":
|
|
|
|
rt.HasCompressOK = true
|
|
|
|
rt.CompressOK, err = strconv.ParseBool(val)
|
|
|
|
if err != nil {
|
|
|
|
return nil, ResumeTokenParsingError
|
|
|
|
}
|
2020-11-07 17:52:10 +01:00
|
|
|
case "embedok":
|
|
|
|
rt.HasEmbedOk = true
|
|
|
|
rt.EmbedOK, err = strconv.ParseBool(val)
|
|
|
|
if err != nil {
|
|
|
|
return nil, ResumeTokenParsingError
|
|
|
|
}
|
|
|
|
case "largeblockok":
|
|
|
|
rt.HasLargeBlockOK = true
|
|
|
|
rt.LargeBlockOK, err = strconv.ParseBool(val)
|
|
|
|
if err != nil {
|
|
|
|
return nil, ResumeTokenParsingError
|
|
|
|
}
|
|
|
|
case "savedok":
|
|
|
|
rt.HasSavedOk = true
|
|
|
|
rt.SavedOk, err = strconv.ParseBool(val)
|
|
|
|
if err != nil {
|
|
|
|
return nil, ResumeTokenParsingError
|
|
|
|
}
|
2018-05-02 21:26:56 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if !rt.HasToGUID {
|
|
|
|
return nil, ResumeTokenDecodingNotSupported
|
|
|
|
}
|
|
|
|
|
|
|
|
return rt, nil
|
|
|
|
|
|
|
|
}
|
2018-06-20 20:20:37 +02:00
|
|
|
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
// if string is empty and err == nil, the feature is not supported
|
|
|
|
func ZFSGetReceiveResumeTokenOrEmptyStringIfNotSupported(ctx context.Context, fs *DatasetPath) (string, error) {
|
|
|
|
if supported, err := ResumeRecvSupported(ctx, fs); err != nil {
|
|
|
|
return "", errors.Wrap(err, "cannot determine zfs recv resume support")
|
|
|
|
} else if !supported {
|
|
|
|
return "", nil
|
|
|
|
}
|
2018-06-20 20:20:37 +02:00
|
|
|
const prop_receive_resume_token = "receive_resume_token"
|
2020-03-27 00:57:33 +01:00
|
|
|
props, err := ZFSGet(ctx, fs, []string{prop_receive_resume_token})
|
2018-06-20 20:20:37 +02:00
|
|
|
if err != nil {
|
|
|
|
return "", err
|
|
|
|
}
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
res := props.Get(prop_receive_resume_token)
|
|
|
|
debug("%q receive_resume_token=%q", fs.ToString(), res)
|
2018-06-20 20:20:37 +02:00
|
|
|
if res == "-" {
|
|
|
|
return "", nil
|
|
|
|
} else {
|
|
|
|
return res, nil
|
|
|
|
}
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
}
|
2018-06-20 20:20:37 +02:00
|
|
|
|
new features: {resumable,encrypted,hold-protected} send-recv, last-received-hold
- **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`
2019-09-11 17:19:17 +02:00
|
|
|
func (t *ResumeToken) ToNameSplit() (fs *DatasetPath, snapName string, err error) {
|
|
|
|
comps := strings.SplitN(t.ToName, "@", 2)
|
|
|
|
if len(comps) != 2 {
|
|
|
|
return nil, "", fmt.Errorf("resume token field `toname` does not contain @: %q", t.ToName)
|
|
|
|
}
|
|
|
|
dp, err := NewDatasetPath(comps[0])
|
|
|
|
if err != nil {
|
|
|
|
return nil, "", errors.Wrap(err, "resume token field `toname` dataset path invalid")
|
|
|
|
}
|
|
|
|
return dp, comps[1], nil
|
2018-06-20 20:20:37 +02:00
|
|
|
}
|