Project

General

Profile

Bug #13837

Monitoring: Google help needed: the stackdriver api doesn't give us the machine type (but it does give us load)

Added by Steven Timm about 4 years ago. Updated almost 3 years ago.

Status:
Closed
Priority:
High
Assignee:
Start date:
09/12/2016
Due date:
10/28/2016
% Done:

90%

Estimated time:
8.00 h
Duration: 47

Description

have to ask onix about this, to see what needs to be done.

History

#1 Updated by Steven Timm almost 4 years ago

Not necessary for the demo but could be nice to have.
Ask Google people if there's anything they can do.
Issue is that we just get relative cpu load from all instance types regardless of size of VM so that 1 core vms and 16 core vms are
weighted the same.

#2 Updated by Steven Timm almost 4 years ago

  • Subject changed from monitoring: the stackdriver api doesn't give us the machine type (but it does give us load) to Google help needed: the stackdriver api doesn't give us the machine type (but it does give us load)

#3 Updated by Steven Timm almost 4 years ago

  • Assignee set to Neha Sharma
  • Priority changed from Normal to High

#4 Updated by Neha Sharma almost 4 years ago

Hi Steve,
Could you please provide more background info and specifics on the issue? Also, what is the correct link for this stackdriver API documentation?

Thanks!

#5 Updated by Steven Timm almost 4 years ago

OK here we go:

the main monitoring code can be checked out of GIT via

git clone ssh://:/cvs/projects/monitoringaccountingbilling

The routine in question can be found in that tree at
trunk/src/monitor-gce/bin/gce_cpumonitor.py

The method in question is get_gce_cpu_usage

It is calling the timeSeries method using v3 of the monitoring API.
This can be seen here:

https://cloud.google.com/monitoring/api/ref_v3/rest/v3/TimeSeries

This returns a data structure separated by instance.

Right now, it does not give us any idea of which machine type each instance is, and how many cores it has.

So for instance if I have 4 core instance at 100% load, and 16 core instance at 25% load, the average load will be (100+25)/2 which is 62.5%. The higher-core instance should be weighted more.
i.e. the right math is (100*4)+(25*16)/(4+16) = 800/20 = 40.

I suggest that you install the code, try to run it, make sure what I have said above is right, then
write up a question that can be submitted to google cloud tech support.'

Steve

#6 Updated by Neha Sharma almost 4 years ago

Reply from Paul (to Burt's email)

"Thanks for this. I'm investigating internally and will attempt to resolve."

#7 Updated by Steven Timm almost 4 years ago

  • Due date set to 11/02/2016
  • Estimated time set to 8.00 h

#8 Updated by Steven Timm almost 4 years ago

  • Subject changed from Google help needed: the stackdriver api doesn't give us the machine type (but it does give us load) to Monitoring: Google help needed: the stackdriver api doesn't give us the machine type (but it does give us load)

#9 Updated by Steven Timm almost 4 years ago

  • Due date changed from 11/02/2016 to 10/28/2016

ST reiterated this question to the google group today in email

#10 Updated by Steven Timm almost 4 years ago

  • Status changed from New to Work in progress
  • % Done changed from 0 to 90

We've had some answers from Solomon and Paul on this but none that make sense yet. Throwing the assignment ticket to Joe Boyd since he has taken
temporary ownership of this code.

#11 Updated by Steven Timm about 3 years ago

  • Status changed from Work in progress to Resolved

This is now considered to be moot as our overall evaluation of the stackdriver API is such that we will not pay the extra fee for it now that they have started charging for it.

#12 Updated by Steven Timm almost 3 years ago

  • Status changed from Resolved to Closed


Also available in: Atom PDF