This is the 2nd installment for the AccountsDb replication.
Summary of Changes
The basic google protocol buffer protocol for replicating updated slots and accounts. tonic/tokio is used for transporting the messages.
The basic framework of the client and server for replicating slots and accounts -- the persisting of accounts in the replica-side will be done at the next PR -- right now -- the accounts are streamed to the replica-node and dumped. Replication for information about Bank is also not done in this PR -- to be addressed in the next PR to limit the change size.
Functionality used by both the client and server side are encapsulated in the replica-lib crate.
There is no impact to the existing validator by default.
Tests:
Observe the confirmed slots replicated to the replica-node.
Observe the accounts for the confirmed slot are received at the replica-node side.
This is the first installment of the AccountsDb replication. Summary of Changes:
The rpc-node for AccountsDb replication executable cherry-picked from Steven's branch
Auto figuring out the snapshot to download via gossip
The replica now can download the snapshot, initialize the bank and AccountsDb, start the JsonRpcService
Integration test of a validator paired with a replica.
* Create solana-poh crate
* Move BigTableUploadService to solana-ledger
* Add solana-rpc to workspace
* Move dependencies to solana-rpc
* Move remaining rpc modules to solana-rpc
* Single use statement solana-poh
* Single use statement solana-rpc
* Add rbpf-cli tool
* Replace explicit file type specification by required positional arg
- the type of file is determined automatically. If ELF magic found,
the program file is considered to be an executable ELF file, and
an assembly input file otherwise.
* Add a note about tool's limitations and input format description
* Add Blockstore protobuf cf type
* Add Rewards message to proto and make generated pub
* Convert Rewards cf to ProtobufColumn
* Add bench
* Adjust tags
* Move solana proto definitions and conversion methods to new crate
* Fix comment and make less pub
* Add account-decoder crate and use to decode vote and system (nonce) accounts
* Update docs
* Rename RpcAccount struct
* s/Rpc/Display
* Call it jsonParsed and update docs
* Revert "s/Rpc/Display"
This reverts commit 6e7149f503f560f1e9237981058ff05642bb7db5.
* s/Rpc/Ui
* Add tests
* Ui more things
* Comments
* Initial commit
* Execute transfers
* Refactor for testing
* Cleanup readme
* Rewrite
* Cleanup
* Cleanup
* Cleanup client
* Use a Null Client to move prints closer to where messages are sent
* Upgrade Solana
* Move core functionality into its own module
* Handle transaction errors
* Merge allocations
* Fixes
* Cleanup readme
* Fix markdown
* Add example input
* Add integration test - currently fails
* Add integration test
* Add metrics
* Use RpcClient in dry-run, just don't send messages
* More metrics
* Fix dry run with no keys
* Only require one approval if fee-payer is the sender keypair
* Fix bugs
* Don't create the transaction log if nothing to put into it;
otherwise the next innvocation won't add the header
* Apply previous transactions to allocations with matching recipients
* Bail out of any account already has a balance
* Polish
* Add new 'balances' command
* 9 decimal places
* Add missing file
* Better dry-run; keypair options now optional
* Change field name from 'bid' to 'accepted'
Also, tolerate precision change from 2 decimal places to 4
* Write to transaction log immediately
* Rename allocations_csv to bids_csv
So that we can bypass bids_csv with an allocations CSV file
* Upgrade Solana
* Remove faucet from integration test
* Cleaner integration test
Won't work until this lands and is released:
https://github.com/solana-labs/solana/pull/9717
* Update README
* Add TravicCI script to build and test (#1)
* Add distribute-stake command (#2)
* Distribute -> DistributeTokens (#3)
* Cache cargo deps (#4)
* Add docs (#5)
* Switch to latest Solana 1.1 release (#7)
* distribute -> distribute-tokens (#9)
* Switch from CSV to a pickledb database (#8)
* Switch from CSV to a pickledb database
* Allow PickleDb errors to bubble up
* Dedup
* Hoist db
* Add finalized field to TransactionInfo
* Don't allow RPC client to resign transactions
* Remove dead code
* Use transport::Result
* Record unconfirmed transaction
* Fix: separate stake account per allocation
* Catch transport errors
* Panic if we attempt to replay a transaction that hasn't been finalized
* Attempt to fix CI
PickleDb isn't calling flush() or close() after writing to files.
No issue on MacOS, but looks racy in CI.
* Revert "Attempt to fix CI"
This reverts commit 1632394f636c54402b3578120e8817dd1660e19b.
* Poll for signature before returning
* Add --sol-for-fees option for stake distributions
* Add --allocations-csv option (#14)
* Add allocations-csv option
* Add tests or GTFO
* Apply review feedback
* apply feedback
* Add read_allocations function
* Update arg_parser.rs
* Fix balances command (#17)
* Fix balances command
* Fix readme
* Add --force to transfer to non-empty accounts (#18)
* Add --no-wait (#16)
* Add ThinClient methods to implement --no-wait
* Plumb --no-wait through
No tests yet
* Check transaction status on startup
* Easier to test
* Wait until transaction is finalized before checking if it failed with an error
It's possible that a minority fork thinks it failed.
* Add unit tests
* Remove dead code and rustfmt
* Don't flush database to file if doing a dry-run
* Continue when transactions not yet finalized (#20)
If those transactions are dropped, the next run will execute them.
* Return the number of confirmations (#21)
* Add read_allocations() unit-test (#22)
Delete the copy-pasted top-level test.
Fixes#19
* Add a CSV printer (#23)
* Remove all the copypasta (#24)
* Move resolve_distribute_stake_args into its own function
* Add stake args to token args
* Unify option names
* Move Command::DistributeStake into DistributeTokens
* Remove process_distribute_stake
* Only unique signers
* Use sender keypair to fund new fee-payer accounts
* Unify distribute_tokens and distribute_stake
* Rename print-database command to transaction-log (#25)
* Send all transactions as quickly as possible, then wait (#26)
* Send all transactions as quickly as possible, then wait
* Exit when finalized or blockhashes have expired
* Don't need blockhash in the CSV output
* Better types
CSV library was choking on Pubkey as a type. PickleDb doesn't have that problem.
* Resend if blockhash has not expired
* Attempt to fix CI
* Move log to stderr
* Add constructor, tuck away client (#30)
* Add constructor, tuck away client
* Fix unwrap() caught by CI
* Fix optional option flagged as required
* Bunch of cleanup (#31)
* Remove untested --no-wait feature
* Make --transactions-db an option, not an arg
So that in the future, we can make it optional
* Remove more untested features
Too many false positives in that santity check. Use --dry-run
instead.
* Add dry-run mode to ThinClient
* Cleaner dry-run
* Make key parameters required
Just don't use them in --dry-run
* Add option to write the transaction log
--dry-run doesn't write to the database. Use this option if you
want a copy of the transaction log before the final run.
* Revert --transaction-log addition
Implement #27 first
* Fix CI
* Update readme
* Fix CI in copypasta
* Sort transaction log by finalized date (#33)
* Make --transaction-db option implicit (#34)
* Move db functionality into its own module (#35)
* Move db functionality into its own module
* Rename tokens module to commands
* Version bump
* Upgrade Solana
* Add solana-tokens to build
* Remove Cargo.lock
* Remove vscode file
* Remove TravisCI build script
* Install solana-tokens
Co-authored-by: Dan Albert <dan@solana.com>
* Separate client types into own crate, so ledger does not need it
Removes about 50 crates of dependency from ledger
* Drop Rpc name from transaction-status types
Lijun Wang
Tyera Eulberg
Jack May
sakridge
Justin Starry
Michael Vines
Alfonso Subiotto Marques
Dmitri Makarov
Solana Maintainers
Jeff Washington (jwash)
Trent Nelson
Greg Fitzgerald
anatoly yakovenko