risingwavelabs/risingwave · error · BatchError

task {:?} not found

Error message

task {:?} not found

What it means

A lookup guard in TaskManager::check_if_task_running: the requested TaskId is absent from the manager's task registry, so the state check fails with 'task {:?} not found'. It fires when a client queries status or output for a task that was never created on this compute node, or whose entry was already removed after completion/cleanup (or after a compute-node restart lost in-memory state). The offending input is a stale or never-registered task_id.

Source

Thrown at src/batch/src/task/task_manager.rs:273

                tracing::trace!("Removed task: {:?}", task.get_task_id());
                // Use `cancel` rather than `abort` here since this is not an error which should be
                // propagated to upstream.
                task.cancel();
                if let Some(heartbeat_join_handle) = task.heartbeat_join_handle() {
                    heartbeat_join_handle.abort();
                }
            }
            None => {
                warn!("Task {:?} not found for cancel", sid)
            }
        };
    }

    /// Returns error if task is not running.
    pub fn check_if_task_running(&self, task_id: &TaskId) -> Result<()> {
        match self.tasks.lock().get(task_id) {
            Some(task) => task.check_if_running(),
            None => bail!("task {:?} not found", task_id),
        }
    }

    pub fn check_if_task_aborted(&self, task_id: &TaskId) -> Result<bool> {
        match self.tasks.lock().get(task_id) {
            Some(task) => task.check_if_aborted(),
            None => bail!("task {:?} not found", task_id),
        }
    }

    #[cfg(test)]
    async fn wait_until_task_aborted(&self, task_id: &TaskId) -> Result<()> {
        use std::time::Duration;
        loop {
            match self.tasks.lock().get(task_id) {
                Some(task) => {
                    let ret = task.check_if_aborted();
                    match ret {

View on GitHub (pinned to 6469eb736d)

Solutions

1. Verify the task was dispatched to this specific compute node; task state is per-node and not shared across the cluster
2. Handle node restarts: in-memory task state is lost on restart, so treat 'not found' as task-unknown and re-derive status from the meta service
3. Check that the task was not already completed and garbage-collected before the status query arrived
4. Ensure the scheduler reads the task status before declaring failure, so completed tasks are recognized rather than reported as missing

When it happens

Trigger: Thrown at src/batch/src/task/task_manager.rs:273 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of risingwavelabs/risingwave@6469eb736d (2026-09-11). Data as JSON: /api/errors/7b08710c1d3ed327. Report an issue: GitHub.