HDDS-6857. [Snapshot] Implement Snapshot Delete CLI and API

[smengcl]

What changes were proposed in this pull request?

• Implement CLI ozone sh snapshot delete.
• Implement Ratis Tx OMSnapshotDeleteRequest.
• Improve CLI output for UX.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-6857

How was this patch tested?

• Add UT TestOMSnapshotDeleteRequest.
• Finish UT TestOMSnapshotDeleteResponse. (Done in HDDS-7862 instead)
• Tested manually in Docker locally:

$ ozone fs -mkdir -p ofs://om/vol1/buck1/dir1/
$ ozone fs -put README.md ofs://om/vol1/buck1/dir1/
$ ozone sh snapshot create /vol1/buck1 snap1
$ ozone sh snapshot delete /vol1/buck1 snap1
$ ozone sh snapshot delete /vol1/buck1 snap1
FILE_NOT_FOUND Snapshot exists but no longer active
$ ozone sh snapshot list /vol1/buck1
[ {
  "volumeName" : "vol1",
  "bucketName" : "buck1",
  "name" : "snap1",
  "creationTime" : 1673583279047,
  "snapshotStatus" : "SNAPSHOT_DELETED",
  "snapshotID" : "08f640e3-95db-4596-9d5f-a47cb7272329",
  "snapshotPath" : "vol1/buck1",
  "checkpointDir" : "-08f640e3-95db-4596-9d5f-a47cb7272329"
} ]

[smengcl]

@GeorgeJahad @hemanthboyina @aswinshakil Please also take a look.

Some of the dead code commented out is intentional in this draft to facilitate discussion. e.g. whether to keep the BUCKET_LOCK or not in OMSnapshotDeleteRequest. They will be cleaned up later.

[neils-dev]

With the snapshot delete cli command is it possible to indicate to the user the amount of space that will be freed with the operation? If not here, then through another method.

$ ozone sh snapshot list /vol1/buck1
[ {
  "volumeName" : "vol1",
  "bucketName" : "buck1",
  "name" : "snap1",
  "creationTime" : 1673583279047,
  "snapshotStatus" : "SNAPSHOT_DELETED",
  "snapshotID" : "08f640e3-95db-4596-9d5f-a47cb7272329",
  "snapshotPath" : "vol1/buck1",
  "checkpointDir" : "-08f640e3-95db-4596-9d5f-a47cb7272329"
} ]

In the example provided, after executing the delete, listing the snapshots (shoiwn above) for the bucket shows the "SNAPSHOT_DELETED" state. With the current example it appears that listing snapshots for the bucket always shows a snapshot in deleted state. We expect it to be removed from the listing at some point? After garbage collection?

[prashantpogde]

Perhaps "Snapshot pending deletion".

[smengcl]

"Snapshot pending deletion" would only apply to SNAPSHOT_DELETED.

I have detailed the snapshot status a bit. Please take another look.

[aswinshakil]

Thanks for working on this @smengcl. I have some minor comments inline.

[smengcl]

With the snapshot delete cli command is it possible to indicate to the user the amount of space that will be freed with the operation? If not here, then through another method.

Hi Neil, we do have snapshot disk usage display planned in HDDS-7744 and has been detailed in the design doc.

$ ozone sh snapshot list /vol1/buck1
[ {
  "volumeName" : "vol1",
  "bucketName" : "buck1",
  "name" : "snap1",
  "creationTime" : 1673583279047,
  "snapshotStatus" : "SNAPSHOT_DELETED",
  "snapshotID" : "08f640e3-95db-4596-9d5f-a47cb7272329",
  "snapshotPath" : "vol1/buck1",
  "checkpointDir" : "-08f640e3-95db-4596-9d5f-a47cb7272329"
} ]

In the example provided, after executing the delete, listing the snapshots (shoiwn above) for the bucket shows the "SNAPSHOT_DELETED" state. With the current example it appears that listing snapshots for the bucket always shows a snapshot in deleted state. We expect it to be removed from the listing at some point? After garbage collection?

It will be cleaned up and removed by SnapshotDeletingTask later, being implemented in HDDS-7740.

[prashantpogde]

LGTM once all the CI tests pass.

[smengcl]

Thanks @prashantpogde for the +1. CI passed. I will merge this to unblock other GC work.

I've filed HDDS-7862 for the TestOMSnapshotDeleteResponse UT TODO.

Thanks @hemantk-12 @aswinshakil @neils-dev for the reviews.