HDDS-6614. EC: Fix Datanode block file INCONSISTENCY during heavy load.

[guihecheng]

What changes were proposed in this pull request?

EC: Fix Datanode block file INCONSISTENCY during heavy load.
Problem descriptions and analysis in the JIRA below.
In brief, we should get the file size by using the fileChannel instead of the file.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-6614

How was this patch tested?

Manual test with experiment programs, seen in the JIRA above.

[kaijchen]

https://docs.oracle.com/javase/7/docs/api/java/io/File.html#length()

public long length()

Returns:
The length, in bytes, of the file denoted by this abstract pathname, or 0L if the file does not exist. Some operating systems may return 0L for pathnames denoting system-dependent entities such as devices or pipes.

[guihecheng]

https://issues.apache.org/jira/browse/HDDS-6614

Yes I've read this, you could read the descriptions in the JIRA and try the provided small program for some hints.

[kerneltime]

This makes sense. Thanks for catching this. LGTM.

[kerneltime]

Minor nit

[umamaheswararao]

Thanks @guihecheng for reporting this. This issue looks very similar to https://issues.apache.org/jira/browse/HDDS-6356
I think it was consistently reproduced in @cchenax environment which was basically starting all DNs in same machine. That time we could not reproduce in actual cluster, so we did not really dig further.

So, to make sure this issue is fixed, how about @cchenax help to reproduce with his env and try this fix there?

[guihecheng]

Thanks @guihecheng for reporting this. This issue looks very similar to https://issues.apache.org/jira/browse/HDDS-6356 I think it was consistently reproduced in @cchenax environment which was basically starting all DNs in same machine. That time we could not reproduce in actual cluster, so we did not really dig further.

So, to make sure this issue is fixed, how about @cchenax help to reproduce with his env and try this fix there?

Oh, yes, may be this problem is similar to that one, but I'm not sure, I'll fix the failed tests and ping cchenax then.

[guihecheng]

Hi @kerneltime @umamaheswararao , thanks for reviewing , test failures addressed and Exception refined.

[cchenax]

Thanks @guihecheng for reporting this. This issue looks very similar to https://issues.apache.org/jira/browse/HDDS-6356 I think it was consistently reproduced in @cchenax environment which was basically starting all DNs in same machine. That time we could not reproduce in actual cluster, so we did not really dig further.

So, to make sure this issue is fixed, how about @cchenax help to reproduce with his env and try this fix there?

ok，I will test it.

[cchenax]

Thanks @guihecheng for reporting this. This issue looks very similar to https://issues.apache.org/jira/browse/HDDS-6356 I think it was consistently reproduced in @cchenax environment which was basically starting all DNs in same machine. That time we could not reproduce in actual cluster, so we did not really dig further.

So, to make sure this issue is fixed, how about @cchenax help to reproduce with his env and try this fix there?

it may be not the same problem.