[BUG] Snapshot % complete calculate is incorrect

[sudosaket]

What is the bug?

When we use the migration assistant to take create a snapshot via console snapshot create, the percent completed stat from the status command are incorrectly computed under the following circumstances:

• a previous snapshot already exists (eg. ma-snapshot-1)
• the snapshot that is being created is an incremental one (eg. ma-snapshot-2)

What are your migration environments?

Opensearch 2.17

How can one reproduce the bug?

• create a snapshot using the migration assistant
• update the snapshot config to change the snapshot name (so that we can take another incremental snapshot)
• create a snapshot using the migration assistant a second time
• check the progress using console snapshot status --deep-check
• the output should be as follows (which is incorrect)

SUCCESS
Snapshot is SUCCESS.
Percent completed: 16.40%
Data GiB done: 56.060/341.817
Total shards: 747
Successful shards: 747
Failed shards: 0
Start time: 2025-07-23 15:07:44
Duration: 0h 2m 58s
Anticipated duration remaining: 0h 15m 12s
Throughput: 320.84 MiB/sec

What is the expected behavior?

The expected behavior should be that the % completion should be 100% as evident from the opensearch APIs:

GET _snapshot/migration_assistant_repo/ma-snapshot-2/_status

# excluding unnecessary fields

{
  "snapshots": [
    {
      "state": "SUCCESS",
      "include_global_state": true,
      "shards_stats": {
        "initializing": 0,
        "started": 0,
        "finalizing": 0,
        "done": 747,
        "failed": 0,
        "aborted": 0,
        "total": 747
      },
      "stats": {
        "incremental": {
          "file_count": 3896,
          "size_in_bytes": 60194325053
        },
        "total": {
          "file_count": 36365,
          "size_in_bytes": 367023410293
        },
        "start_time_in_millis": 1753283264535,
        "time_in_millis": 178921
      }
    }
  ]
}

Do you have any additional context?

We should compute the progress from the incremental stats if present.

[jugal-chauhan]

Hi @sudosaket, thank you for reporting this and for the clear reproduction steps!

You’re right, this is a great catch. When dealing with incremental snapshots, the current progress computation in console snapshot status --deep-check doesn’t account for the consecutive nature of the operation. As you’ve pointed out, although the OpenSearch APIs show the snapshot has completed successfully with all shards done, the Migration Assistant CLI still reports incomplete progress due to relying on total snapshot size instead of incremental stats.

I’ve also added a dedicated task within our internal Jira Epic to fix this specific progress reporting issue by switching to the incremental snapshot statistics when available.

Having said that, it is going to be a lower priority fix at this point because user experience is going to change soon with the Delta Snapshots update.

[sudosaket]

Sounds good! This update is scheduled for within 3 months?

[peternied]

@sudosaket Thanks for creating this issue, there could definitely be improvements here. If you wanted to see an update sooner, what do you think about making a contribution? Here is a [link] to where the logic for the snapshot status message is defined.

[sudosaket]

Okay, let me take a look at this!