Must have sensors of YDB SDK

[asmyasnikov]

PR to https://github.com/ydb-platform/ydb-go-sdk-metrics

Current metrics > 160 and histogram buckets > 600 Its too much. Need to refactor adapter ydb-go-sdk-metrics for reduce number of metrics

• [ ] table:
  • [ ] pool:
    • [x] limit
    • [x] size
    • [x] inflight
    • [ ] idle
    • [ ] create_in_progress
    • [x] queue depth (wait session from pool)
    • [x] latency of session inflight usage
    • [x] latency of wait session from pool
  • [x] sessions:
    • [x] alive by node id
• [ ] query:
  • [ ] limit
  • [ ] size
  • [ ] inflight
  • [ ] idle
  • [ ] create_in_progress
  • [ ] queue depth (wait session from pool)
  • [ ] latency of session inflight usage
  • [ ] latency of wait session from pool
  • [ ] sessions:
    • [ ] alive by node id
• [ ] database/sql:
  • [x] conns (inflight)
  • [x] txs (inflight)
• [ ] requests:
  • [x] status codes counter (Ok or error by type)
  • [ ] inflight (or counter?) by node id / endpoint
  • [x] discovery requests counter
  • [x] endpoints count by dc
  • [x] pessimized endpoints
  • [ ] latency of table/ExecuteDataQuery
  • [x] TLI rate
• [ ] retryer:
  • [x] attempts count
  • [x] intermediate error codes
• [ ] topic: https://github.com/ydb-platform/ydb-go-sdk/issues/840
  • [ ] pool size
  • [ ] message time in pool
  • [ ] retry attempts
  • [ ] retry intermediate error codes
  • [ ] counter of read messages
  • [ ] counter of write messages
  • [ ] write message ack time
  • [ ] uptime of stream

[asmyasnikov]

1. table - trace.Table
2. requests - trace.Driver
3. retryer - trace.Retry + trace.Table.OnDo + trace.TableOnDoTx
4. topic - trace.Topic