Diagnostics Reporting

AttentionThis page documents an earlier version. Go to the latest (v2.1)version.

yb-master and yb-tserver binaries report cluster diagnostics to Yugabyte’s diagnostics service every time a new cluster gets created and every hour thereafter. User data stored in YugabyteDB as well as any personally identifiable information regarding the user is never collected or reported.

Data collected

The data collected depends on the collection level set. See the section below on how to change the collection level.

Collection level: low

  1. Collection time
  2. Cluster uuid
  3. Node uuid
  4. Node type (tserver or master)
  5. Number of masters (yb-master processes)
  6. Number of tablet servers (yb-tserver processes)
  7. Number of tables
  8. Number of tablets
  9. Cluster configuration (gflags)

Collection level: medium [Default]

  1. Everything we collect for level “low” plus:
  2. Performance metrics
  3. RPC metrics

Collection level: high

Same as medium level. In other words, no extra information is collected for this level. It will be used in the future to collect additional dianostics information such as error logs.

Example of metrics collected

  1. [
  2.     {
  3.         "type": "tablet",
  4.         "id": "eca8b3cfb2ee4eca94dde519634e1e38",
  5.         "attributes": {
  6.             "partition": "hash_split: [16380, 19110)",
  7.             "table_name": "redis",
  8.             "table_id": "82f5f7ab81a44923b5f544fbd0664afe"
  9.         },
  10.         "metrics": [
  11.             {
  12.                 "name": "log_reader_bytes_read",
  13.                 "value": 0
  14.             },
  15.             {
  16.                 "name": "log_reader_entries_read",
  17.                 "value": 0
  18.             },
  19.             {
  20.                 "name": "log_reader_read_batch_latency",
  21.                 "total_count": 0,
  22.                 "min": 0,
  23.                 "mean": 0,
  24.                 "percentile_75": 0,
  25.                 "percentile_95": 0,
  26.                 "percentile_99": 0,
  27.                 "percentile_99_9": 0,
  28.                 "percentile_99_99": 0,
  29.                 "max": 0,
  30.                 "total_sum": 0
  31.             }
  32.          ]
  33.     }
  34. }

Example of RPCs being collected

  1. {
  2.     "inbound_connections": [
  3.         {
  4.             "remote_ip": "10.150.0.20:41134",
  5.             "state": "OPEN",
  6.             "remote_user_credentials": "{real_user=yugabyte, eff_user=}",
  7.             "processed_call_count": 2456
  8.         },
  9.         {
  10.             "remote_ip": "10.150.0.20:54141",
  11.             "state": "OPEN",
  12.             "remote_user_credentials": "{real_user=yugabyte, eff_user=}",
  13.             "processed_call_count": 2471
  14.         }
  15.     ]
  16. }

Configuration flags

You can add the following flags while starting the yb-master and yb-tserver binaries to configure the diagnostics reporting behavior of YugabyteDB.

FlagDefaultDescription
—callhome_collection_levellowCollection level with possible values of low, medium, or high
—callhome_interval_secs3600Collection interval in seconds
—callhome_urlhttp://diagnostics.yugabyte.comEndpoint where diagnostics information is reported
—callhome_enabledtrueControls whether diagnostics information is collected and reported. Set to false to disable collection.