ADBDEV-6011: Fix restore helper hides actual err

[RekGRpth]

Fix restore helper hides actual err

The commit 4371a94 moved the LoopEnd label so that the err variable began to be
overwritten and it will no longer be used below. To fix it, this patch renames
the err variable returned by flushAndCloseRestoreWriter to errPipe.

The integration test has been adapted to error-free behavior.

---

Fix restore helper hides actual err

The commit 4371a94 moved the LoopEnd label so that the err

gpbackup/helper/restore_helper.go

Lines 308 to 310 in 9b07afe

	err = errors.Wrap(err, strings.Trim(errBuf.String(), "\x00"))
	} else {
	err = errors.Wrap(err, "Error copying data")

error began to be overwritten

gpbackup/helper/restore_helper.go

Line 322 in 9b07afe

err = flushAndCloseRestoreWriter(currentPipe, tableOid)

and it will no longer be used below

gpbackup/helper/restore_helper.go

Line 335 in 9b07afe

if err != nil {

To fix it, this patch renames err variable returned by flushAndCloseRestoreWriter to errPipe.

[dkovalev1]

Looks logical, but what is the scenario when it does matter? Can we reproduce the problem fixed and make a test?

[RekGRpth]

Looks logical, but what is the scenario when it does matter? Can we reproduce the problem fixed and make a test?

Problem was found theoretically

[red1452]

Change the commit message:

The commit https://github.com/arenadata/gpbackup/commit/4371a94305a7a77ce3e6990a9691a3986d9426d9 moved the LoopEnd label so that the err variable began to be
overwritten and it will no longer be used below. To fix it, this patch renames
the err variable returned by flushAndCloseRestoreWriter to errPipe.

[RekGRpth]

Change the commit message:

The commit https://github.com/arenadata/gpbackup/commit/4371a94305a7a77ce3e6990a9691a3986d9426d9 moved the LoopEnd label so that the err variable began to be
overwritten and it will no longer be used below. To fix it, this patch renames
the err variable returned by flushAndCloseRestoreWriter to errPipe.

changed

[dkovalev1]

Looks logical, but what is the scenario when it does matter? Can we reproduce the problem fixed and make a test?

Problem was found theoretically

Can you at least explain when the problem happens and which cases to patch solves? If it's not possible to supply a test.

[RekGRpth]

Looks logical, but what is the scenario when it does matter? Can we reproduce the problem fixed and make a test?

Problem was found theoretically

Can you at least explain when the problem happens and which cases to patch solves? If it's not possible to supply a test.

test was added

The base branch was changed.

[dkovalev1]

This change seems logical and allows gpbackup to report error. However, the test system is far from parfect. Tests are very difficult to read, trace and understand.