Monitorare cluster Big Data usando azdata e kubectl
Questo articolo illustra come visualizzare lo stato di un cluster Big Data usando azdata e kubectl.
Importante
Il componente aggiuntivo per i cluster Big Data di Microsoft SQL Server 2019 verrà ritirato. Il supporto per i cluster Big Data di SQL Server 2019 terminerà il 28 febbraio 2025. Tutti gli utenti esistenti di SQL Server 2019 con Software Assurance saranno completamente supportati nella piattaforma e fino a quel momento il software continuerà a ricevere aggiornamenti cumulativi di SQL Server. Per altre informazioni, vedere il post di blog relativo all'annuncio e Opzioni per i Big Data nella piattaforma Microsoft SQL Server.
Usare azdata
È anche possibile usare i comandi di azdata per visualizzare sia gli endpoint che lo stato del cluster.
Endpoint di servizio
Eseguire l'autenticazione al cluster Big Data con azdata login. Impostare il parametro
--controller-endpointsull'indirizzo IP esterno dell'endpoint controller.azdata login --endpoint https://<ip-address-of-controller-svc-external>:30080 --username <user-name>Specificare il nome utente e la password configurati per il controller (AZDATA_USERNAME e AZDATA_PASSWORD) durante la distribuzione.
Per l'autenticazione di Active Directory, il comando è:
azdata login --endpoint https://<control_domain_name>:30080 --auth adEseguire
azdata bdc endpoint listper ottenere un elenco con una descrizione di ogni endpoint e i relativi valori di porta e indirizzo IP.azdata bdc endpoint list -o tableL'elenco seguente mostra un output di esempio di questo comando:
Description Endpoint Ip Name Port Protocol ------------------------------------------------------ --------------------------------------------------------- -------------- ----------------- ------ ---------- Gateway to access HDFS files, Spark https://11.111.111.111:30443 11.111.111.111 gateway 30443 https Spark Jobs Management and Monitoring Dashboard https://11.111.111.111:30443/gateway/default/sparkhistory 11.111.111.111 spark-history 30443 https Spark Diagnostics and Monitoring Dashboard https://11.111.111.111:30443/gateway/default/yarn 11.111.111.111 yarn-ui 30443 https Application Proxy https://11.111.111.111:30778 11.111.111.111 app-proxy 30778 https Management Proxy https://11.111.111.111:30777 11.111.111.111 mgmtproxy 30777 https Log Search Dashboard https://11.111.111.111:30777/kibana 11.111.111.111 logsui 30777 https Metrics Dashboard https://11.111.111.111:30777/grafana 11.111.111.111 metricsui 30777 https Cluster Management Service https://11.111.111.111:30080 11.111.111.111 controller 30080 https SQL Server Master Instance Front-End 11.111.111.111,31433 11.111.111.111 sql-server-master 31433 tcp HDFS File System Proxy https://11.111.111.111:30443/gateway/default/webhdfs/v1 11.111.111.111 webhdfs 30443 https Proxy for running Spark statements, jobs, applications https://11.111.111.111:30443/gateway/default/livy/v1 11.111.111.111 livy 30443 https
Visualizzare lo stato del cluster
È possibile visualizzare lo stato del cluster con il comando azdata bdc status show.
azdata bdc status show
Suggerimento
Per eseguire il comando per lo stato, è prima di tutto necessario accedere con il comando azdata login, mostrato nella sezione precedente sugli endpoint.
Di seguito viene mostrato un output di esempio di questo comando:
Bdc: ready Health Status: healthy
===========================================================================================================================================================================================================================================
Services: ready Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Servicename State Healthstatus Details
spark ready healthy -
sql ready healthy -
hdfs ready healthy -
control ready healthy -
gateway ready healthy -
app ready healthy -
Spark Services: ready Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Resourcename State Healthstatus Details
sparkhead ready healthy StatefulSet sparkhead is healthy
storage-0 ready healthy StatefulSet storage-0 is healthy
Sql Services: ready Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Resourcename State Healthstatus Details
master ready healthy StatefulSet master is healthy
compute-0 ready healthy StatefulSet compute-0 is healthy
data-0 ready healthy StatefulSet data-0 is healthy
storage-0 ready healthy StatefulSet storage-0 is healthy
Hdfs Services: ready Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Resourcename State Healthstatus Details
nmnode-0 ready healthy StatefulSet nmnode-0 is healthy
zookeeper ready healthy StatefulSet zookeeper is healthy
storage-0 ready healthy StatefulSet storage-0 is healthy
sparkhead ready healthy StatefulSet sparkhead is healthy
Control Services: ready Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Resourcename State Healthstatus Details
controldb ready healthy StatefulSet controldb is healthy
control ready healthy ReplicaSet control is healthy
metricsdc ready healthy DaemonSet metricsdc is healthy
metricsui ready healthy ReplicaSet metricsui is healthy
metricsdb ready healthy StatefulSet metricsdb is healthy
logsui ready healthy ReplicaSet logsui is healthy
logsdb ready healthy StatefulSet logsdb is healthy
mgmtproxy ready healthy ReplicaSet mgmtproxy is healthy
controlwd ready healthy ReplicaSet controlwd is healthy
Gateway Services: ready Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Resourcename State Healthstatus Details
gateway ready healthy StatefulSet gateway is healthy
App Services: ready Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Resourcename State Healthstatus Details
appproxy ready healthy ReplicaSet appproxy is healthy
Stato di risorse specifiche
È possibile visualizzare lo stato di una risorsa specifica nel cluster con il comando azdata bdc status show. Quando si usa questo comando è possibile filtrare usando il parametro --resource. Ecco alcuni esempi di input per il parametro --resource:
- master
- control
- compute-0
- storage-0
- gateway
Il comando seguente visualizza ad esempio lo stato del pool di archiviazione:
azdata bdc status show --all --resource storage-0
Per visualizzare lo stato di tutti i componenti che eseguono un servizio specifico, è necessario usare il gruppo di comandi corrispondente azdata bdc <serviceName> status show. Ad esempio:
azdata bdc sql status show --allazdata bdc hdfs status show --allazdata bdc spark status show --all
Output di esempio:
Storage-0: ready Health Status: healthy
===========================================================================================================================================================================================================================================
Instances: running Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Instancename State Healthstatus Details
storage-0-0 running healthy Pod storage-0-0 is healthy
storage-0-1 running healthy Pod storage-0-1 is healthy
Dashboards
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Name Url
nodeMetricsUrl https://13.91.50.9:30777/api/v1/bdc/instances/storage-0-1/status/nodemetrics/ui
sqlMetricsUrl https://13.91.50.9:30777/api/v1/bdc/instances/storage-0-1/status/sqlmetrics/ui
logsUrl https://13.91.50.9:30777/api/v1/bdc/instances/storage-0-1/status/logs/ui
Suggerimento
Eseguire il comando relativo allo stato con i parametri --all per informazioni aggiuntive sull'integrità, inclusi collegamenti ai dashboard di metriche e log corrispondenti all'istanza specifica. Di seguito è riportato un esempio di output quando vengono usati i parametri --all:
Spark: ready Health Status: healthy
===========================================================================================================================================================================================================================================
Resources: ready Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Resourcename State Healthstatus Details
sparkhead ready healthy StatefulSet sparkhead is healthy
storage-0 ready healthy StatefulSet storage-0 is healthy
Sparkhead Resources: running Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Instancename State Healthstatus Details
sparkhead-0 running healthy Pod sparkhead-0 is healthy
sparkhead-1 running healthy Pod sparkhead-1 is healthy
Dashboards
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Name Url
nodeMetricsUrl https://13.91.50.9:30777/api/v1/bdc/instances/sparkhead-1/status/nodemetrics/ui
sqlMetricsUrl https://13.91.50.9:30777/api/v1/bdc/instances/sparkhead-1/status/sqlmetrics/ui
logsUrl https://13.91.50.9:30777/api/v1/bdc/instances/sparkhead-1/status/logs/ui
Storage-0 Resources: running Health Status: healthy
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Instancename State Healthstatus Details
storage-0-0 running healthy Pod storage-0-0 is healthy
storage-0-1 running healthy Pod storage-0-1 is healthy
Dashboards
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Name Url
nodeMetricsUrl https://13.91.50.9:30777/api/v1/bdc/instances/storage-0-1/status/nodemetrics/ui
sqlMetricsUrl https://13.91.50.9:30777/api/v1/bdc/instances/storage-0-1/status/sqlmetrics/ui
logsUrl https://13.91.50.9:30777/api/v1/bdc/instances/storage-0-1/status/logs/ui
Visualizzare lo stato del controller
È possibile visualizzare lo stato del controller con il comando azdata bdc control status show. Questo comando fornisce collegamenti simili ai dashboard di monitoraggio correlati ai componenti del controller del cluster Big Data.