[AMBARI-25998] Host uuid is not getting synced to other collectors, causing NPE while accessing metric

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: metrics-3.0.0
• Fix Version/s: 2.8.0, 2.7.8
• Component/s: ambari-metrics
• Labels: pull-request-available
• Assignee: Mohammad Arshad
• Reporter: Mohammad Arshad
• Created: 29/Aug/23 09:09
• Updated: 07/Nov/23 17:33
• Resolved: 07/Nov/23 17:33

Description

Host uuid is not getting synced with other collectors, causing NPE while accessing metric

java.lang.NullPointerException
        at org.apache.ambari.metrics.core.timeline.PhoenixHBaseAccessor.getLastTimelineMetricFromResultSet(PhoenixHBaseAccessor.java:457)
        at org.apache.ambari.metrics.core.timeline.PhoenixHBaseAccessor.getLatestMetricRecords(PhoenixHBaseAccessor.java:1150)
        at org.apache.ambari.metrics.core.timeline.PhoenixHBaseAccessor.getMetricRecords(PhoenixHBaseAccessor.java:966)
        at org.apache.ambari.metrics.core.timeline.HBaseTimelineMetricsService.getTimelineMetrics(HBaseTimelineMetricsService.java:294)
        at org.apache.ambari.metrics.webapp.TimelineWebServices.getTimelineMetrics(TimelineWebServices.java:267)

Steps to reproduce the issue

1. Push below metrics through collector-1

{
  "metrics": [
    {
      "metricname": "cpu1",
      "appid": "app1",
      "hostname": "host1",
      "starttime": 1693294119309,
      "metrics": {
        "1693294119309": 100,
        "1693294129309": 200
      }
    }
  ]
}

After metadata sync interval elapsed, performed following operations
2. GET http://COLLECTOR-1-HOST:6188/ws/v1/timeline/metrics/hosts returned

{
    "host1": [
        "app1"
    ],
...
}

3. GET http://COLLECTOR-2-HOST:6188/ws/v1/timeline/metrics/hosts returned

{ "host1": [ "app1" ], ... }

4. GET http://COLLECTOR-1-HOST:6188/ws/v1/timeline/metrics?metricNames=cpu1&appId=app1&hostname=host1 returned
{
"metrics": [
{
"metricname": "cpu1",
"appid": "app1",
"hostname": "host1",
"timestamp": 0,
"starttime": 1693294119309,
"metrics":

{ "1693294129309": 200.0 }

,
"metadata": {}
}
]
}

5. GET http://COLLECTOR-2-HOST:6188/ws/v1/timeline/metrics?metricNames=cpu1&appId=app1&hostname=host1 returned

{
    "exception": "NullPointerException",
    "javaClassName": "java.lang.NullPointerException"
}

When host name is being synced, host id also should be synced to fix this issue.
Similar to this issue another issue https://issues.apache.org/jira/browse/AMBARI-25963 was raised, but that NPE was because metrics uuid was not synced, this issue is because of host uuid is not synced.

Issue Links

• links to: GitHub PR#3750
• links to: GitHub Pull Request #124