[Disk manager] correctly handle deletion of non-existent disk; improve test on concurrent overlay disk creation and deletion

[gy2411]

In #4339 there was a bug. This bug leads to a leaking base disk slot and, consequenly, to hanging retire base disk task.

Scenario:

• Overlay disk is creating
• Concurrenlty, this overlay disk is deleting
• Disk is deleted successfully
• CreateOverlayDiskTask successfully acquires base disk slot
• CreateOverlayDiskTask sees that disk is deleted and starts to cancel
• On cancel, CreateOverlayDiskTask sees that the disk is deleted and does not proceed with releasing base disk

In this pr we make the following changes:

• Task does not stop after storage.DeleteDisk call even if the is is already deleted.
• Improve the test using CheckBaseDiskSlotReleased.
• Increase DeletedDiskExpirationTimeout parameter in test config. Otherwise, the deleted disk gets cleaned up from the database too quickly and the test does not work.

Without the fix, the improve test fails with error:

common.go:711: 
                Error Trace:    /-S/cloud/disk_manager/internal/pkg/facade/testcommon/common.go:711
                                                        /-S/cloud/disk_manager/internal/pkg/facade/disk_service_test/disk_service_test.go:1273
                Error:          Received unexpected error:
                                Non retriable error, Silent=false: internal inconsistency: slot &{overlayDiskID:TestDiskServiceDeleteDiskWhenCreationIsInFlightAAAA overlayDiskKind:0 overlayDiskSize:134217728 baseDiskID:base-6c1b5f58-9f2f-44fb-8f27-6856e31de178 imageID:TestDiskServiceDeleteDiskWhenCreationIsInFlightAAAA zoneID:zone-a allottedSlots:1 allottedUnits:1 releasedAt:{wall:0 ext:62135596800 loc:0x1fe8200} targetZoneID: targetBaseDiskID: targetAllottedSlots:0 targetAllottedUnits:0 generation:0 status:0} should be released

[github-actions]

Note

This is an automated comment that will be appended during run.

🔴 linux-x86_64-relwithdebinfo: some tests FAILED for commit dabd103.

TESTS	PASSED	ERRORS	FAILED	FAILED BUILD	SKIPPED	MUTED?
1255	1253	0	2	0	0	0

🔴 linux-x86_64-relwithdebinfo: some tests FAILED for commit dabd103.

TESTS	PASSED	ERRORS	FAILED	FAILED BUILD	SKIPPED	MUTED?
47	46	0	1	0	0	0

🟢 linux-x86_64-relwithdebinfo: all tests PASSED for commit dabd103.

TESTS	PASSED	ERRORS	FAILED	FAILED BUILD	SKIPPED	MUTED?
45	45	0	0	0	0	0

[github-actions]

Note

This is an automated comment that will be appended during run.

🟢 linux-x86_64-relwithdebinfo: all tests PASSED for commit 0838a21.

TESTS	PASSED	ERRORS	FAILED	FAILED BUILD	SKIPPED	MUTED?
1255	1255	0	0	0	0	0

[gy2411]

Нужно продолжать удаление диска, даже если диска нет в сторадже.

В частности, таск удаления диска, отсутсвтующего в сторадже, должен отправить в NBS запрос на удаление этого диска.

Поэтому при пустой ZoneID не делаем return, а забираем зону из запроса (как мы это и делали до того, как посадили баг: https://github.com/ydb-platform/nbs/pull/4339/files#diff-1605caf3e850a302b598eb7fed17f3d13cae8a4defd1b6c2f18df892bcbf8060)

[gy2411]

Выделил эту правку в отдельный pr #4910

[gy2411]

И всё же лучше это место вернуть, как было.

[github-actions]

Note

This is an automated comment that will be appended during run.

🟢 linux-x86_64-relwithdebinfo: all tests PASSED for commit 0473505.

TESTS	PASSED	ERRORS	FAILED	FAILED BUILD	SKIPPED	MUTED?
1255	1255	0	0	0	0	0

[github-actions]

Note

This is an automated comment that will be appended during run.

🟢 linux-x86_64-relwithdebinfo: all tests PASSED for commit d97e0cd.

TESTS	PASSED	ERRORS	FAILED	FAILED BUILD	SKIPPED	MUTED?
1255	1255	0	0	0	0	0

[gy2411]

We don't want to proceed deletion of disk when disk does not exist. So another approach is proposed, see #5539.

[gy2411]

Close in favour to #5539.