Internal error: Non Panic Task error: task 113 was cancelled. This was likely caused by a bug in DataFusion's code and we would welcome that you file an bug report in our issue tracker

[chenquan]

datafusion/datafusion/physical-plan/src/stream.rs

Lines 126 to 131 in 05bb569

	// This should only occur if the task is
	// cancelled, which would only occur if
	// the JoinSet were aborted, which in turn
	// would imply that the receiver has been
	// dropped and this code is not running
	return Some(internal_err!("Non Panic Task error: {e}"));

Hello, I don't understand why this error occurs?

[alamb]

In what context did you see the error happen?

Was it during shutdown of the tokio runtime?

Was a thread panicing?

[chenquan]

@alamb Reproduction steps:

1. clone

git clone https://github.com/chenquan/rsflow

2. switch working directory

cd rsflow

3. run

cargo run --package rsflow --bin rsflow  -- --config ./examples/generate_example.yaml

4. Send ctl-c signa, then an error occurred.

where the error occurred:

https://github.com/chenquan/rsflow/blob/85714766f5af131da5001bc96882ba0f6a18843f/src/processor/sql.rs#L60-L63

[chenquan]

Was it during shutdown of the tokio runtime?

send a off signa.

Was a thread panicing?

no.

[chenquan]

@alamb Hello, have you made any progress?

[alamb]

@alamb Hello, have you made any progress?

Hi @chenquan -- I am not actively working on this issue and don't think I will have any plans to do so.

Given you have a clear reproducer I think the assumption in the code that this is an unexpected error condition is not true

In other words I think we should update the error to be a runtime error (that the task was cancelled) a

I would be happy to review a PR to fix the issue

[chenquan]

@alamb Hello, have you made any progress?

Hi @chenquan -- I am not actively working on this issue and don't think I will have any plans to do so.

Given you have a clear reproducer I think the assumption in the code that this is an unexpected error condition is not true

In other words I think we should update the error to be a runtime error (that the task was cancelled) a

I would be happy to review a PR to fix the issue

@alamb Hello, thank you very much! In addition, I want to know how to control the cancellation of tasks by myself. Currently, I have observed that the program automatically cancels the task after receiving the off signal.

[alamb]

@alamb Hello, thank you very much! In addition, I want to know how to control the cancellation of tasks by myself. Currently, I have observed that the program automatically cancels the task after receiving the off signal.

it probably depends on what rsflow is doing (I am not familiar with that program)

Basically if something starts shutting down the tokio Runtime that is going to cancel all outstanding tasks. Perhaps you need to modify rsflow to catch SIGINT and initiate a clean shutdown / wait for outstanding jobs to be done

[chenquan]

@alamb Hello, thank you very much! In addition, I want to know how to control the cancellation of tasks by myself. Currently, I have observed that the program automatically cancels the task after receiving the off signal.

it probably depends on what rsflow is doing (I am not familiar with that program)

Basically if something starts shutting down the tokio Runtime that is going to cancel all outstanding tasks. Perhaps you need to modify rsflow to catch SIGINT and initiate a clean shutdown / wait for outstanding jobs to be done

@alamb

Thanks again, I think I've found the reason.

[alamb]

Thanks again, I think I've found the reason.

Out of curiosity, what did you find?

[alamb]

I also filed a ticket to fix the error type

• "Non Panic Task error" is not an internal error #15108

[chenquan]

Thanks again, I think I've found the reason.

Out of curiosity, what did you find?

Before the main thread exits, I do not exit the tokio runtime where the datafusion is located normally.