celeborn/METRICS.md

Metrics

We provide various metrics about memory, disk, and important procedures. These metrics could help identify performance issue or monitor RSS cluster.

Prerequisites

1.Enable RSS metrics. set rss.metrics.system.enabled = true
2.You need to install prometheus(https://prometheus.io/)
We provide an example for prometheus config file

# prometheus example config
global:
  scrape_interval: 15s
  evaluation_interval: 15s

scrape_configs:
  - job_name: "RSS"
    metrics_path: /metrics/prometheus
    scrape_interval: 15s
    static_configs:
      - targets: [ "master-ip:9098","worker1-ip:9096","worker2-ip:9096","worker3-ip:9096","worker4-ip:9096" ]

3.You need to install Grafana server(https://grafana.com/grafana/download)

4.Import RSS dashboard into grafana. You can find RSS dashboard at assets/grafana/rss-dashboard.json.

Optional

We recommend you to install node exporter (https://github.com/prometheus/node_exporter) on every host, and configure prometheus to scrape information about the host. Grafana will need a dashboard (dashboard id:8919) to display host details.

global:
  scrape_interval: 15s
  evaluation_interval: 15s

scrape_configs:
  - job_name: "RSS"
    metrics_path: /metrics/prometheus
    scrape_interval: 15s
    static_configs:
      - targets: [ "master-ip:9098","worker1-ip:9096","worker2-ip:9096","worker3-ip:9096","worker4-ip:9096" ]
  - job_name: "node"
    static_configs:
      - targets: [ "master-ip:9100","worker1-ip:9100","worker2-ip:9100","worker3-ip:9100","worker4-ip:9100" ]

Here is an example of grafana dashboard importing.

Details

MetricName	Role	Description
WorkerCount	master	The count of active workers.
BlacklistedWorkerCount	master	The count of workers in blacklist.
OfferSlotsTime	master	The time of offer slots.
PartitionSize	master	The estimated partition size of last 20 flush window whose length is 15 seconds by defaults.
RegisteredShuffleCount	master and worker	The value means count of registered shuffle.
CommitFilesTime	worker	CommitFiles means flush and close a shuffle partition file.
ReserveSlotsTime	worker	ReserveSlots means acquire a disk buffer and record partition location.
FlushDataTime	worker	FlushData means flush a disk buffer to disk.
OpenStreamTime	worker	OpenStream means read a shuffle file and send client about chunks size and stream index.
FetchChunkTime	worker	FetchChunk means read a chunk from a shuffle file and send to client.
MasterPushDataTime	worker	MasterPushData means handle pushdata of master partition location.
SlavePushDataTime	worker	MasterPushData means handle pushdata of slave partition location.
PushDataFailCount	worker	The count of failed PushData or PushMergedData.
TakeBufferTime	worker	TakeBuffer means get a disk buffer from disk flusher.
SlotsAllocated	worker	Slots allocated in last hour
NettyMemory	worker	The value measures all kinds of transport memory used by netty.
SortTime	worker	SortTime measures the time used by sorting a shuffle file.
SortMemory	worker	SortMemory means total reserved memory for sorting shuffle files .
SortingFiles	worker	This value means the count of sorting shuffle files.
DiskBuffer	worker	Disk buffers are part of netty used memory, means data need to write to disk but haven't been written to disk.
PausePushData	worker	PausePushData means the count of worker stopped receiving data from client.
PausePushDataAndReplicate	worker	PausePushDataAndReplicate means the count of worker stopped receiving data from client and other workers.

Implementation

RSS master metric : com/aliyun/emr/rss/service/deploy/master/MasterSource.scala
RSS worker metric : com/aliyun/emr/rss/service/deploy/worker/WorkerSource.scala and com.aliyun.emr.rss.common.metrics.source.NetWorkSource

Grafana Dashboard

We provide a grafana dashboard for RSS Grafana-Dashboard. The dashboard was generated by grafana of version 8.5.0. Here are some snapshots: