HDDS-15764. [Recon] Show error card for unavailable datanode in Cluster Capacity

[priyeshkaratha]

What changes were proposed in this pull request?

On the Recon Cluster Capacity page, the datanode detail card seeds its default selection to the first datanode in the filtered list, regardless of whether that DN is reachable. When the default-selected DN is unavailable — its pending-deletion query failed, signalled by the pendingBlockSize === -1 sentinel — the card still rendered used / remaining / committed values from the storage report. For an offline or unreachable DN these values can be stale, so the UI silently displayed outdated capacity data. The node-selector dropdown already disables such DNs (disabledOpts), but the default seeding (hostNames[0]) can still land on one.

Now it will show error card when all DNs are down, otherwise if we selected DN which is not available it will auto switch to one of the available DN

Note : Code is generated using AI

What is the link to the Apache JIRA

HDDS-15764

How was this patch tested?

Tested manually and updated unit test case

[chihsuan]

Thanks @priyeshkaratha LGTM. Tested locally as well.

[devmadhuu]

@priyeshkaratha , changes largely LGTM, just a point worth noting that, it is always better to point the default to the first available DN and only fall back to the error card when every DN is unavailable, else user always lands on an error card on first load and must manually switch — even though a healthy DN exists.

[priyeshkaratha]

@priyeshkaratha , changes largely LGTM, just a point worth noting that, it is always better to point the default to the first available DN and only fall back to the error card when every DN is unavailable, else user always lands on an error card on first load and must manually switch — even though a healthy DN exists.

Thanks @devmadhuu. This make sense. I have updated PR which will do auto switch and if all of the DNs are down it will show error card.

[devmadhuu]

Thanks @priyeshkaratha for improving the patch. LGTM +1

[priyeshkaratha]

Thanks @devmadhuu and @chihsuan for reviewing the patch.