HDDS-6203. CleanUp incomplete gz files during Container move

[Xushaohong]

What changes were proposed in this pull request?

If the target temporary directory is small, the concurrent downloading threads will have a contest for space and leave incomplete files. These should be cleaned up.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-6203

How was this patch tested?

manual tests

[JacksonYao287]

Suggested change

	LOG.error("Failed to write the stream buffer to {} for container {}",
	outputPath, containerId, e);
	deleteOutputOnFailure();
	onError(e);

[guihecheng]

IMO, the original LOG msg is better, because by looking at the JIRA descriptions, actually the download via network is successful, the failure happens at sync the downloaded data into the on-disk file.
So we don't call onError() here which should be called on network transfer faliures, and it may not be a good style to call an overrided method in another(onError in onNext here).

[guihecheng]

May be we could extract common error handling stuff into a new method.

[JacksonYao287]

May be we could extract common error handling stuff into a new method.

looks good. i agree

[ChenSammi]

stream.close(); should be called together with deleteOutputOnFailure();

[JacksonYao287]

thanks @Xushaohong for the work! the idea looks good!
if an IOException is thrown out when writing to file here, we should :
1 close the file stream
2 delete the incomplete downloaded file/
3 close this downloading connection

all the above is impletemented in onError, so seems better to just call onError directly here

[ChenSammi]

The last patch LGTM, +1.

Thanks @Xushaohong for the contribution, and @guihecheng @JacksonYao287 for the code review.