Fixing the in memory raw shard hash update.
Passing the seq_no of index update to verify that the index records are consistent.
Restart rw session after completing last primary shard sync.
Fixed issue in fs read write by multiple threads.
Hpfs log sync modifications.
- Introduced input hash which can later be used to query the ledger.
- Returned input hash and ledger info at input submission.
- Updated js client lib to support input hash.
- Updated consensus proposal candidate expiration rules.
* Hpfs index file is updated to match the file with the hpfs version.
* Added updated hpfs binary.
* Version log changes because Plog is not initialized at the time of version init.
* full history announcement in peer challenge and ability to send random messages only to full history nodes.
* log sync infrastructure codes.
* Resolving merge conflict induced errors.
* Fork detection implementation.
* Cleanup.
* Resolving PR comments.
* hpfs binary update, hpfs log serving and persisting
* Fixed merge conflict and added missing log
Co-authored-by: Savinda Senevirathne <savindadilsara@gmail.com>
* Fixed issue in consensus contract output sorting.
* Improved load test script.
* Removed flatbuffer verifier restrictions to support large no. of users and inputs.
* Improved consensus proposal cleanup.
* Restored the earlier proposal cleanup logic.
* Minor fixes.
* const fix.
Basic infrastructure for log file sync.
A fork is detected by comparing the hpfs log index file against the ledger.
Min and max ledger sequence hashes are sent to a full history node to request log history.
- Truncation is invoked by making truncate system call on /hpfs::index.<seq_no> file.
- Updating hpfs binary.
- Truncation waits until any active RW and RO sessions are finished.
* Return empty hash when set or vector is empty in crypto.
* Changing <operator in sequence_hash struct.
* Fixing vm script and allowing only single sync target per vpath.
* Making vm cluster log level to debug.
* Increasing hpws max message size and fixing lcl mode in vm cluster scripts.
* lcl block fix.
* Remove and request historical shards at the startup
* Shard history requesting only at the first consensus round
* Removed test log
* Skip max shard seq no file in removing loop
* Updated the code comments
* Persisting condition changed
* Fixed code comment typos
* Fixed code comment typos
* Resolved PR comments
* Halt consensus until completing only the latest shard sync
* Added meaningful comments
* Resolved PR comments and updated hpfs binary
* Logic enhancement and cleanup
* Cleanup the code comment
* Skip proposal and npl messages from non unl nodes
* Minor code comment change
* Resolved PR comments and Changed weakly connected status check
* Changed weakly connected status counter
* Added a comment
- Added new config for max input ledger offset inside contract section.
- Updated implementation of getting and updating patch config inside node and c contracts.
- Skip inputs if max ledger offset exceeds, when handling user inputs and nup messages.
* Adding full history mode and max shard configs for primary and blob shards.
* Commenting validation checks for hpfs responses until bug is fixed.
* Comment update.
- The ledgers are stored in a sqlite database in ledger file system.
- Ledgers are organized in shards inside primary folder.
- Raw inputs are saved as shards inside blob folder. No input files are created if neither inputs nor outputs are available.
- Both last primary shard hash and last blob shard hashes are subjected to consensus and necessary sync operations are performed if out of sync.
- Hpfs sync support setting single sync targets from both ends of the list. (Targets set from front are prioritized).
- Contract and ledger syncs and serving are performed independently.
