fix: improve weight setting reliability and synchronization #342
Conversation
- Prevent concurrent weight setting attempts between base neuron and validator\n- Use consistent timing reference for weight updates\n- Fix unnecessary f-string in logging\n- Add proper locking in validator scoring loop\n- Wait for transaction finalization to prevent ancient block errors
|
PR description is too short and seems to not fulfill PR template, please fill in |
|
PR description is too short and seems to not fulfill PR template, please fill in |
|
PR description is too short and seems to not fulfill PR template, please fill in |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #342 +/- ##
==========================================
- Coverage 62.67% 61.83% -0.85%
==========================================
Files 23 23
Lines 1321 1344 +23
==========================================
+ Hits 828 831 +3
- Misses 493 513 +20 ☔ View full report in Codecov by Sentry. |
- Change non-critical errors to warnings (trending queries, subnet config, protocol API, tweet requests) - Keep appropriate warning levels for validation failures - Files modified: masa/validator/forwarder.py, masa/base/validator.py, masa/miner/twitter/tweets.py
|
PR description is too short and seems to not fulfill PR template, please fill in |
- Change non-critical connection errors from ERROR to WARNING/DEBUG level - Add summary counts instead of individual errors for miner responses - Show unique miner versions instead of full list - Improve error handling and response validation - Add concise status summaries for ping and volume operations
|
PR description is too short and seems to not fulfill PR template, please fill in |
- Replace full UID lists with summary counts in logging - Update get_random_miner_uids and get_uncalled_miner_uids to show concise stats - Remove redundant logging of individual UIDs - Clean up imports and fix linter errors
|
PR description is too short and seems to not fulfill PR template, please fill in |
- Add missing score_miners method to fix UnboundLocalError\n- Initialize weights_rate_limit at startup\n- Add proper locking to prevent concurrent scoring\n- Improve error handling and logging in scoring methods\n- Maintain consistent scoring intervals based on weights_rate_limit
- Change non-critical connection errors to warnings\n- Add summary counts for successful responses\n- Improve error handling for connection failures\n- Clean up logging messages for better clarity
|
PR description is too short and seems to not fulfill PR template, please fill in |
…r handling - Restored tweet validation, score tracking, and tweet export logic - Maintained improved error handling and logging - Added success/failure counts for better monitoring - Improved clarity of log messages
|
PR description is too short and seems to not fulfill PR template, please fill in |
…le preserving scoring logic - Split large methods into focused units, change non-critical errors to WARNING level, add error statistics and summary counts, improve error recovery while preserving all scoring logic
|
PR description is too short and seems to not fulfill PR template, please fill in |
|
PR description is too short and seems to not fulfill PR template, please fill in |
…mple runners in scripts/
🧪 Testing
The changes have been tested in both development and production environments, with particular attention to:
💥 Impact
These changes significantly improve the validator's reliability and maintainability by:
None. All changes maintain backward compatibility while improving existing functionality.
🔜 Next Steps