tip | title | description | author | discussions-to | status | type | layer | created |
---|---|---|---|---|---|---|---|---|
9 |
Local Snapshot File Format |
File format to export/import ledger state |
Luca Moser (@luca-moser) <[email protected]> |
Active |
Standards |
Interface |
2020-08-25 |
This RFC defines a file format for local snapshots which is compatible with Chrysalis Phase 2.
Nodes create local snapshots to produce ledger representations at a point in time of a given milestone to be able to:
- Start up from a recent milestone instead of having to synchronize from the genesis transaction.
- Delete old transaction data below a given milestone.
Current node implementations use a local snapshot file format which only works with account based ledgers. For Chrysalis Phase 2, this file format has to be assimilated to support a UTXO based ledger.
Since a UTXO based ledger is much larger in size, this RFC proposes two formats for snapshot files:
- A
full
format which represents a complete ledger state. - A
delta
format which only contains diffs (created and consumed outputs) of milestones from a given milestone index onwards.
This separation allows nodes to swiftly create new delta snapshot files, which then can be distributed with a companion full snapshot file to reconstruct a recent state.
Unlike the current format, these new formats do not include spent addresses since this information is no longer held by nodes.
All types are serialized in little-endian
A full ledger snapshot file contains the UTXOs (outputs
section) of a node's confirmed
milestone (ledger_milestone_index
). The diffs
contain the diffs to rollback the outputs
state to regain the ledger state of the snapshot milestone at (seps_milestone_index
).
While the node producing such a full ledger state snapshot could theoretically pre-compute the actual snapshot milestone state, this is deferred to the consumer of the data to speed up local snapshot creation.
A delta ledger state local snapshot only contains the diffs
of milestones starting from a
given ledger_milestone_index
. A node consuming such data must know the state of the ledger at ledger_milestone_index
.
Defines an output.
Name | Type | Description | ||||||||||||||||||||||||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Message Hash | Array[32] | The hash of the message in which the transaction was contained which generated this output. | ||||||||||||||||||||||||||||||||||||||||||
Transaction hash | Array[32] | The hash of the transaction which generated this output. | ||||||||||||||||||||||||||||||||||||||||||
Output index | uint16 | The index of this output within the transaction. | ||||||||||||||||||||||||||||||||||||||||||
Output oneOf |
SigLockedSingleDeposit
SigLockedDustAllowanceDeposit
|
Defines the diff a milestone produced by listing the created/consumed outputs and the milestone payload itself.
Name | Type | Description | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Milestone Payload length | uint32 | Denotes the length of the milestone payload. | |||||||||
Milestone Payload | Array[Milestone Payload length] | The milestone payload in its serialized binary form. | |||||||||
Treasury Input
only included if milestone contains a receipt |
|
||||||||||
Created Outputs Count | uint64 | The amount of outputs generated with this milestone diff. | |||||||||
Created Outputs anyOf |
Output |
||||||||||
Consumed Outputs Count | uint64 | The amount of outputs consumed with this milestone diff. | |||||||||
Consumed Outputs anyOf |
Output |
Defines what a full snapshot file contains.
Name | Type | Description |
Version | byte | Denotes the version of this file format. |
Type | byte | Denotes the type of this file format. Value 0 denotes a full snapshot. |
Timestamp | uint64 | The UNIX timestamp in seconds of when this snapshot was produced. |
Network ID | uint64 | The ID of the network to which this snapshot is compatible. |
SEPs milestone index | uint64 | The milestone index for which the SEPs were generated for. |
Ledger milestone index | uint64 | The milestone index of which the UTXOs within the snapshot are from. |
SEPs count | uint64 | The amount of SEPs contained within this snapshot. |
Outputs count | uint64 | The amount of UTXOs contained within this snapshot. |
Milestone diffs count | uint64 | The amount of milestone diffs contained within this snapshot. |
Treasury Output Milestone Hash | Array[32] | The milestone hash of the milestone which generated the treasury output. |
Treasury Output Amount | uint64 | The amount of funds residing on the treasury output. |
SEPs |
SEP Array[32] |
|
Outputs |
Output |
|
Milestone Diffs |
Milestone Diff |
Defines what a delta snapshot contains.
Name | Type | Description |
Version | byte | Denotes the version of this file format. |
Type | byte | Denotes the type of this file format. Value 1 denotes a delta snapshot. |
Timestamp | uint64 | The UNIX timestamp in seconds of when this snapshot was produced. |
Network ID | uint64 | The ID of the network to which this snapshot is compatible. |
SEPs milestone index | uint64 | The milestone index for which the SEPs were generated for. |
Ledger milestone index | uint64 | The milestone index up on which this delta snapshot builts up from. |
SEPs count | uint64 | The amount of SEPs contained within this snapshot. |
Milestone diffs count | uint64 | The amount of milestone diffs contained within this snapshot. |
SEPs |
SEP Array[32] |
|
Milestone Diffs |
Milestone Diff |
Nodes need to support this new format.
- In conjunction with a companion full snapshot, a tool or node can "truncate" the data from a delta snapshot back to a single full snapshot. In that case, the
ledger_milestone_index
andseps_milestone_index
would be the same. In the example above, given the full and delta snapshots, one could produce a new full snapshot for milestone 1350. - Since snapshots may include millions of UTXOs, code generating such files needs to stream data directly onto disk instead of keeping the entire representation in memory. In order to facilitate this, the count denotations for SEPs, UTXOs and diffs are at the beginning of the file. This allows code generating snapshot files to only have to seek back once after the actual count of elements is known.
- Is all the information to startup a node from the local snapshot available with the described format?
Copyright and related rights waived via CC0.