- Agent now requests testruns only from primary server (first in list)
- Results are submitted to all configured servers in parallel
- Secondary servers accept external testruns via new v2 endpoint
- Added auto-creation of gateway and testrun records on secondary servers
- New database queries: get_or_create_gateway, insert_external_testrun
- Client library enhanced with submit_results_with_context method
Agents can now connect to multiple APIs and randomly select one for each testrun:
- Accept multiple --server arguments in format "address:port:auth_key"
- Randomly shuffle server list before attempting connections
- Try each server until a testrun is obtained
- Submit results back only to the API that provided the testrun
- Continue to next server if one is down or has no testruns available
Implement dual database support for SQLite and PostgreSQL through Cargo feature flags.
The implementation uses a query wrapper that automatically converts SQLite-style ?
placeholders to PostgreSQL-style $1, $2, ... placeholders at runtime.
Key changes:
- Add query wrapper functions that handle placeholder conversion
- Convert all sqlx::query\! macros to use wrapper functions
- Handle type conversions between databases (i64 vs i32)
- Add feature-gated implementations for database-specific SQL syntax
- Update Makefile with clippy targets for both database features
- Document database support in README
- Created query_wrapper module with functions to automatically convert
SQLite ? placeholders to PostgreSQL $1, $2, ... format
- Updated build.rs to handle mutually exclusive feature flags
- Modified one query in mixnodes.rs as proof of concept
- Added type conversions for PostgreSQL compatibility (u32->i64, u16->i32)
This is a checkpoint commit before converting all queries to use the wrapper.
* renamed nym-api config fields
* decouple rewarder startup from network monitor
* additional sections in nym-api config
* removed vesting queries in circulating supply calculator
* added memoized field for last submitted performance measurement
* wip: performance contract refresher
* cleaned up various contract caches
* modified cache refresher to allow passing update fn
* implement performance cache refreshing
* updated lefthook.yml to run cargo fmt
* impl NodePerformanceProvider trait
* dynamically using specific performance provider
* pre warm up performance contract cache and forbid the mode if its empty
* clippy
* introduce fallback setting for performance contract if value for given epoch is not available
* move some functions around
* initialised basic structure for the performance contract
* shared code for contract testing
* unified common testing methods between performance and nym pool contracts
* impl of ExecuteMsg for the contract
* impl of QueryMsg for the contract
* setting initial authorised NMs during instantiation
* additional tests and fixes
* ibid
* scaffolding for client traits
* completed client traits
* clippy
* naive add performance contract to testnet manager
* placeholder values for the performance contract address
* introduced admin messages to purge old measurements from the storage
* introduced check ensuring performance data is only added to bonded nodes
* wip - dvpn directory cache
* Endpoint & cache
* /gateways works
- SkimmedNode data still missing
- need to move probe models to monorepo
* Rest of the data for /gateways
* Revert before merge: pin deps to cheddar release
* Filter gw by country
* Return percent string instead of u8
* Filter by semver
* Bump package version
* Fix probe types
* Reorg
* Add exit, entry endpoints
* Different entry/exit selection criteria
* Date fix migration
* Unpin from cheddar
* Revert "Unpin from cheddar"
This reverts commit f17239075b.
* Validation with celes
* PR feedback
* Fix path
* Bump version
---------
Co-authored-by: Mark Sinclair <mmsinclair@users.noreply.github.com>
- keep all the internal-dev wasm pieces as future examples
- everything previously was going to be removed
- shows functioning wasm interaction with the js
durch