What causes "Error resetting workflow state: {}" in conductor-oss/conductor?

Check the wrapped exception — typically an index connectivity/timeout issue.

conductor-oss/conductor · error · TransientException

Error resetting workflow state: {}

Error message

Error resetting workflow state: {}

What it means

Thrown by resetWorkflow when the indexDAO.asyncRemoveWorkflow / removeWorkflow call inside its try block raises any Exception. resetWorkflow first verifies the workflow exists (getWorkflowModelFromDataStore) and removes it from the ExecutionDAO, then attempts to clear it from the index; an index failure at that step aborts with this TransientException.

Source

Thrown at core/src/main/java/com/netflix/conductor/core/dal/ExecutionDAOFacade.java:487

    }

    /**
     * Reset the workflow state by removing from the {@link ExecutionDAO} and removing this workflow
     * from the {@link IndexDAO}.
     *
     * @param workflowId the workflow id to be reset
     */
    public void resetWorkflow(String workflowId) {
        getWorkflowModelFromDataStore(workflowId, true);
        executionDAO.removeWorkflow(workflowId);
        try {
            if (properties.isAsyncIndexingEnabled()) {
                indexDAO.asyncRemoveWorkflow(workflowId);
            } else {
                indexDAO.removeWorkflow(workflowId);
            }
        } catch (Exception e) {
            throw new TransientException("Error resetting workflow state: " + workflowId, e);
        }
    }

    public List<TaskModel> createTasks(List<TaskModel> tasks) {
        tasks.forEach(this::externalizeTaskData);
        return executionDAO.createTasks(tasks);
    }

    public List<Task> getTasksForWorkflow(String workflowId) {
        return getTaskModelsForWorkflow(workflowId).stream()
                .map(TaskModel::toTask)
                .collect(Collectors.toList());
    }

    public List<TaskModel> getTaskModelsForWorkflow(String workflowId) {
        return executionDAO.getTasksForWorkflow(workflowId);
    }

View on GitHub (pinned to cf7c3e4a8a)

Solutions

Check the wrapped exception — typically an index connectivity/timeout issue.
Because the DAO record was already removed, manually remove the orphaned index document for that workflowId if consistency matters.
Retry the reset once the index is healthy; or run a separate index-cleanup.
Ensure the indexDAO backend has capacity for the delete load.

Defensive patterns

Strategy: retry

Try / catch

try {
    executionDAOFacade.resetWorkflow(workflowId);
} catch (TransientException e) {
    // DAO record already removed; index may be orphaned -> retry/cleanup index
}

Prevention

After a resetWorkflow TransientException, plan to reconcile the index separately (DAO is already clean).
Ensure the indexDAO backend is healthy before batch resets.
Wrap reset operations in a retry-with-backoff that tolerates partial completion.

When it happens

Trigger: Calling resetWorkflow(workflowId) when the ExecutionDAO removal succeeded but the subsequent index removal threw. Note the DAO record is already gone at this point, so the index is left with a dangling entry.

Common situations: Index (Elasticsearch) unavailable during a reset/restart workflow operation. A 'restart workflow' API call that resets state and then re-runs. Index overloaded or rejecting deletes.

Related errors

AI-assisted analysis of conductor-oss/conductor@cf7c3e4a8a (2026-08-14). Data as JSON: /api/errors/9f95cdc9dd872a3b. Report an issue: GitHub.