- Introduced README.md for the file sync issue after gRPC migration, outlining the problem, root cause, and proposed solutions.
- Added release notes for Crawlab 0.7.0 highlighting community features and improvements.
- Created a README.md for the specs directory to provide an overview and usage instructions for LeanSpec.
Ensure MustGetSortOption returns a default bson.D{{"_id", -1}} on parse errors and
GetPaginationPipeline uses {_id: -1} when sort is nil to provide consistent default ordering.
Implement createWorkerClient to return a workerTaskClient when a node has an active gRPC stream.
Add workerTaskClient that wraps the server.TaskServiceServer, stubs other client RPCs and forwards
CheckProcess calls directly to server.CheckProcess for querying worker process status.
Configure server-side keepalive (EnforcementPolicy and ServerParameters) to align with client settings and prevent connection timeouts after network disconnection/reconnection.
Optimistically mark node as online/active and persist ActiveAt when a node
subscribes, logging success or warning on failure. Send a node notification
in Pro mode if the status changed. Also tidy import ordering.
Ensure the reconnecting flag is reset on failed attempts so subsequent retries can proceed,
and explicitly trigger a reconnection attempt after the backoff period to keep retrying recovery.
- add maxReconnectionWait and reconnectionCheckInterval constants for reconnection readiness polling
- introduce waitForFullReconnectionReady() to verify: connection READY, clients registered, and ability to obtain critical service clients (model/task) within short timeouts
- ensure reconnecting flag is cleared immediately on reconnection failure and only cleared after full readiness checks on success
- improve logging around reconnection stabilization and readiness checks
- add RetryWithBackoff helper to grpc client for exponential retry with backoff and reconnection-aware handling
- increase reconnectionClientTimeout to 90s and introduce connectionStabilizationDelay; wait briefly after reconnection to avoid immediate flapping
- refresh reconnection flag while waiting for client registration and improve cancellation message
- replace direct heartbeat RPC with RetryWithBackoff in WorkerService (use extended timeout)
- use RetryWithBackoff for worker node status updates in task handler and propagate errors
Add svc.getGrpcClient() helper and use it when obtaining TaskClient so task fetch and
subscribe operations don't hold a stale client instance after ResetGrpcClient().
- Trigger reconnection proactively from Get*WithTimeout when underlying connection is in
SHUTDOWN or TRANSIENT_FAILURE to avoid returning stale/unusable clients.
- Add debug/info logs around client registration, connection attempts, closing existing
connections, connection initiation, reconnection start, backoff retry and successful
reconnection (including current state and registration status).
- Surface more context in reconnection and connection logs to aid diagnostics.
- master: add TIMING logs in setWorkerNodeOnline to mark start and completed DB update
- handler: log node status for reconnection debugging and include active/enabled values in "node not active or enabled" error
- Downgraded protoc-gen-go-grpc and protoc versions for compatibility.
- Added CheckProcess method to TaskService with corresponding request and response types.
- Updated Subscribe and Connect methods to use new generic client stream types.
- Refactored server and client implementations for Subscribe and Connect methods.
- Ensured backward compatibility by maintaining existing method signatures where applicable.
- Added necessary handler for CheckProcess in the service descriptor.
