Modified Endpoints for multiple managers and new compaction summary

[dlmarion]

Closes #6247

[dlmarion]

@DomGarguilo - this changes two endpoints. Let me know if you want any changes in the response object formats and I can make them.

[DomGarguilo]

@DomGarguilo - this changes two endpoints. Let me know if you want any changes in the response object formats and I can make them.

I think with these changes the JSON will be something like:

  {
    "type=MANAGER;group=default;address=localhost:9995": [
      { "...": "metric..." },
      { "...": "metric..." }
    ]
  }

That will work but is not quite as nice to parse. I'm thinking it might be better to send a list of objects instead of a map. so something like this:

  [
    {
      "server": {
        "type": "MANAGER",
        "resourceGroup": "default",
        "address": "localhost:9995"
      },
      "metrics": [
        { "...": "metric..." },
        { "...": "metric..." }
      ]
    }
  ]

Not exactly sure what the code will look like but could create something like record ServerIdView(String type, String resourceGroup, String address) for reuse in the other endpoints that need the ServerId too

[keith-turner]

Tried running these changes w/ #6217 and only saw primary manager metrics. Have not tested it yet, but #6304 may fix this by making this code pull from all manager processes.

[keith-turner]

Would probably be easier on the front end code to roll the metrics up. Could do something like the following.

  public record CompactionTableSummary(String tableId, long running){}
  public record CompactionGroupSummary(String groupId, long running, long queued, long rejected/*, other metrics */){}
  public record CompactionSummary(List<CompactionTableSummary> tableSummaries, List<CompactionGroupSummary> groupSummaries){}

[keith-turner]

Maybe the list above should be maps.

[keith-turner]

When I look at the json output for this I am seeing _ replaced w/ . sometimes. Below are two examples of this. This is code is probably causing that.

  "tableCompactions": {
    "1": 6
  },
  "groupCompactions": {
    "ci_small": 3,
    "ci_lrg": 3
  },

        {
          "name": "accumulo.compaction.queue.jobs.queued",
          "type": "GAUGE",
          "tags": [
            {
              "queue.id": "ci.small"
            },
            {
              "statistic": "value"
            }
          ],
          "value": 0
        }

[dlmarion]

This is code is probably causing that.

Looks right. I'm surprised we are not using that method in more places. I'll create a ticket for it. For now, suggest not using , _, or - in a queue name as the metric data won't line up.