change StreamLoadException from NonFatal to Fatal

[zhaorongsheng]

Proposed changes

Issue Number: close #205

Problem Summary:

Throw exception when commit unsuccessfully. And change the exception from NonFatal to Fatal exception to make spark application throw Fatal exception and make the job failed eventually.

Checklist(Required)

1. Does it affect the original behavior: (Yes)
2. Has unit tests been added: (No Need)
3. Has document been added or modified: (No Need)
4. Does it need to update dependencies: (No)
5. Are there any changes that cannot be rolled back: (No)

Further comments

If this is a relatively large or complex change, kick off the discussion at dev@doris.apache.org by explaining why you chose the solution you did and what alternatives you considered, etc...

[JNSimba]

Here we should just check whether the status of commit and abort is success?

[zhaorongsheng]

Here we should just check whether the status of commit and abort is success?

Spark class ListenerBus( line 127) will not throw exception when it check the exception is NonFatal exception. So we should change the exception to Fatal exception.

[zhaorongsheng]

It's a supplementary PR of #206

[JNSimba]

Here we should just check whether the status of commit and abort is success?

Spark class ListenerBus( line 127) will not throw exception when it check the exception is NonFatal exception. So we should change the exception to Fatal exception.

Does this mean that even if commit throws an exception, the spark program will not exit?

[zhaorongsheng]

Here we should just check whether the status of commit and abort is success?

Spark class ListenerBus( line 127) will not throw exception when it check the exception is NonFatal exception. So we should change the exception to Fatal exception.

Does this mean that even if commit throws an exception, the spark program will not exit?

Yep. Spark ListenerBus will check whether the exception is fatal or not . If the exception is not fatal the spark program only prints some warn logs and the status of the program is success.

[JNSimba]

@zhaorongsheng Thanks for reply, But I tried it locally, after commit throws an exception, it will go to abort, and then the spark task will stop. Can you provide a case that can be reproduced?