Query results and handle failures - BlockX Docs

Track the task identifier together with its EVM chain, input block or bundle, and target table. Those values let you connect execution results to the data you intended to produce.

Keep the running connection

After submission, keep WatchTasks attached to the selected Worker. The stream sends a current snapshot and then the terminal update. Reconnect within the configured grace period after an interruption. GetTaskResult can inspect a result but does not keep a running task attached.

Read the terminal result

Check	Action
Task is still `RUNNING`	Continue watching
Task is `SUCCEEDED`	Inspect the selected Writer Plugin result
Task is `FAILED`	Read the failure code, message, and retryability
Task is unknown	Verify the Worker and task identifier; consider result retention

For ReturnValueHandler, read the returned outputs. For table writes, retain the writer statistics and any job_id, then confirm storage readiness separately.

Decide whether to repeat work

Distinguish an RPC or connection failure from a returned task failure.
Query the original task when the submission outcome is uncertain.
Fix input, function, or configuration errors before repeating.
Inspect the existing write state if the task reached its Writer Plugin.
Create new work only after deciding that another execution is appropriate.

A repeated identifier on the same Worker can return the existing result during its retention window. It does not force the failed task to execute again. A new identifier can produce another write.

Locate the failure

Stage	Inspect
Input	Source table, block or bundle availability, and configuration
Function	Input fields, explicit EVM context, audit findings, and contract errors
Output	Target schema, Writer Plugin configuration, and write-job state
Connection	Proxy, selected Worker, and Watch lifetime

Continue reading