ambari-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Massimilian Mattetti" <MASSI...@il.ibm.com>
Subject Re: how to interpret the host metrics retrieved from Rest API
Date Wed, 15 Feb 2017 11:18:53 GMT
I was collecting the metrics pulling from the address 
http://ambari-server:8080/api/v1/clusters/test-cluster/hosts?fields=Hosts/last_heartbeat_time,metrics

every 10 seconds. All the CPU, Memory and Network metrics were fine but 
the disk throughput was most of the time 0 (calculated by doing 
[read_bytes(t1+10s) - read_bytes(t1)] / 10s). Now I started to use the 
Amabari Collector API and it looks better since it has the 
"read_bps,write_bps" metrics. As you say, I can see from the collector 
that all the metrics are updated every 10 seconds, is that a parameter 
that can be configured? What is the retention time for those metrics (I am 
not sure if one of these is the right one 
https://docs.hortonworks.com/HDPDocuments/Ambari-2.2.1.1/bk_ambari_reference_guide/content/ams_aggregated_ttl_settings.html
)?
Thanks.

Regards,
Max




From:   Aravindan Vijayan <avijayan@hortonworks.com>
To:     "user@ambari.apache.org" <user@ambari.apache.org>
Date:   10/02/2017 21:38
Subject:        Re: how to interpret the host metrics retrieved from Rest 
API



Hi Massimilian,

Ideally, the disk metrics should be collected every 10 seconds as well. 
The granularity of the data depends on the time range requested as well as 
whether the hostname query parameter is passed or not. 

Can you share the exact GET call that you used to fetch disk, cpu, memory 
and network metrics from AMS?

-- 
Thanks and Regards,
Aravindan Vijayan

From: Massimilian Mattetti
Reply-To: "user@ambari.apache.org"
Date: Thursday, February 9, 2017 at 9:45 AM
To: "user@ambari.apache.org"
Subject: Re: how to interpret the host metrics retrieved from Rest API

I managed to calculate the Disk throughput, but I still have the problem 
of getting stale data. I noticed that the values of the disk metrics 
change once per minutes, while the values for cpu, memory and network are 
constantly updated. Is there a way to overcome this problem?
Thanks. 

Best Regards,
Max



From:        Aravindan Vijayan <avijayan@hortonworks.com>
To:        "user@ambari.apache.org" <user@ambari.apache.org>
Date:        08/02/2017 19:12
Subject:        Re: how to interpret the host metrics retrieved from Rest 
API



Hi Massimilian,

Ambari Metrics Service uses psutil library to collect system metrics. The 
metrics's explanation can be found here - https://pythonhosted.org/psutil/
. 

In case there is no metric in psutil with a given name in AMS, it might 
have been re-named. Please refer to the following file for renaming - 
https://github.com/apache/ambari/blob/trunk/ambari-metrics/ambari-metrics-host-monitoring/src/main/python/core/host_info.py
. 

-- 
Thanks and Regards,
Aravindan Vijayan

From: Massimilian Mattetti
Reply-To: "user@ambari.apache.org"
Date: Wednesday, February 8, 2017 at 12:41 AM
To: "user@ambari.apache.org"
Subject: how to interpret the host metrics retrieved from Rest API

Hi all,

I am trying to monitor the resource usage of a cluster by pulling the 
metrics from the rest API of Ambari (Version 2.4.2.0). I am sending 
request with the following path:
http://ambari-server:8080/api/v1/clusters/test/hosts?fields=Hosts/last_heartbeat_time,metrics


There are few metrics that are really hard to understand. For instance, 
the metrics of the disk:

"disk" : {
         "disk_free" : 3278.83,
         "disk_total" : 4948.57,
         "read_bytes" : 5.0036226048E10,
         "read_count" : 197716.0,
         "read_time" : 8.7998397E7,
         "write_bytes" : 1.0003064832E10,
         "write_count" : 822311.0,
         "write_time" : 1.6088341E7
       },

How should I interpret the read_time? Why the values take minutes to 
change (I am pulling every 10 seconds) even if the host is under high disk 
I/O? Is there a legend that explains the meaning of each one of the 
metrics?
Thanks.

Best Regards,
Max







Mime
View raw message