HDDS-14960. OM Web UI dashboard for Ozone Snapshot

[jojochuang]

What changes were proposed in this pull request?

HDDS-14960. OM Web UI dashboard for Ozone Snapshot

Please describe your PR in detail:

• Create a Ozone Snapshot dashboard tab in OM Web UI, that displays:

1. Usage statstics – number of active, deleted snapshots, size of current snapshot cache.
2. List of snapshot diff jobs – status text, progress, creation time; search&sort capabilities.
3. Snapshot internal metrics in OmSnapshotInternalMetrics.

The implementation includes:

1. Frontend code to display usage statistics, snapshot diff jobs, internal metrics, code to support search and sort capabilities.
2. Backend code: SnapshotDiffManager implements SnapshotDiffManagerMXBean and register it with the runtime.
3. Test code to ensure the MXBean is registered correctly, and the values is expected.

SnapshotDiffJob getCodec() renamed to codec() to avoid clashes with JMXBean conventions.

Gemini CLI model gemini-3-flash-preview

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-14960

How was this patch tested?

Unit tests, manually inspected frontend web UI after executing this command (screen shot attached)

ozone sh volume create vol1
ozone sh bucket create vol1/bucket1
ozone sh snapshot create vol1/bucket1 snap1
ozone sh key put vol1/bucket1/key1 /etc/krb5.conf
ozone sh snapshot create vol1/bucket1 snap2
ozone sh snapshot diff vol1/bucket1 snap1 snap2
ozone sh snapshot lsDiff vol1/bucket1 -a
ozone sh key delete vol1/bucket1/key1
ozone sh snapshot create vol1/bucket1 snap3
ozone sh snapshot diff vol1/bucket1 snap2 snap3
ozone sh snapshot diff vol1/bucket1 snap1 snap3
ozone sh snapshot lsDiff vol1/bucket1 -a```

[sadanand48]

Thanks @jojochuang for the patch, One more metric is the time taken to compute the diff. This is not part of the SnapDiffJob class currently , we can add it in a separate jira as well

[sadanand48]

These metrics capture the num of active/deleted snapshots created since the last restart of OM. I think we should indicate that accordingly

[errose28]

Hi @jojochuang I don't think we should duplicate metrics on the web UIs. The page will become overloaded with too much information which Grafana can display better. I suggest making a Grafana dashboard for snapdiff (which AI can also do easily) which contains Usage statstics, Snapshot internal metrics, and any other relevant metrics.

The list of snapdiff jobs is categorical information so that makes sense to put in the Web UI for now, since Recon is not pulling this information.

[jojochuang]

don't think we should duplicate metrics on the web UIs. The page will become overloaded with too much information which Grafana can display better. I suggest making a Grafana dashboard for snapdiff (which AI can also do easily) which contains Usage statstics, Snapshot internal metrics, and any other relevant metrics.

The list of snapdiff jobs is categorical information so that makes sense to put in the Web UI for now, since Recon is not pulling this information.

I'll keep the list of snapshot diff jobs. We don't yet have the Grafana dashboards tracking Snapshot internal metrics, which is why I put it here initially. That being said, these internal metrics aren't that self explanatory. I would like to make a page that integrates these metrics and present in a way that helps administrators understand, because the internals of Snapshots are asynchronous.

Also reminded me there are a few snapshot related metrics within DeletingServiceMetrics that should be included.

[jojochuang]

Updated. This is the current screenshot.

[jojochuang]

https://issues.apache.org/jira/browse/HDDS-13181 added the OM Snapshot internal metrics to Grafana.

It doesn't include numSnapshotSetProperties, numSnapshotSetPropertyFails, and no defragmentation metrics.

[smengcl]

Is this a subset of #10055 ? Or should this be merged first before that one. Looks to be the latter

[smengcl]

On a side note, the search box does not quite work as one would expect. It only filters the currently displayed entries, not the entire set. At least this is the case on SCM Web UI Node Status section as I tried earlier. Filed HDDS-15007

[jojochuang]

Is this a subset of #10055 ? Or should this be merged first before that one. Looks to be the latter

Yes I want to implement these in steps.