976851 - [monitoring-story] dashboard for account sign in and create time by service type filtered by user agent, region

Status: RESOLVED FIXED

(Cloud Services :: Server: Firefox Accounts, defect)

Description

[Edwin Wong [:edwong]]

2. As a SRE/QA manager, I want account create or sign in times by service tier(db, crypto, network).
Acceptance criteria:
-  SRE/QA manager has a dashboard showing create and sign ins time for custom time spans that can be filtered by  browser, OS, and region broken out by each tier of service.
- An email alert is sent when create or sign in exceeds a set threshold.
Desired Timeframe: Fx29 Deploy

Comment 1

[Katie Parlante]

Edwin: here's a first pass, just measures auth server endpoints: https://kibana.fxa.us-west-2.prod.mozaws.net/index.html#/dashboard/elasticsearch/Request%20Times

Comment 2

[Katie Parlante]

(In reply to Ben Wong [:mostlygeek] from comment #3)
> FYI: 
> 
> This same functionality is available on stackdriver: 
> 
> https://app.stackdriver.com/services/load-balancers/fxa-p-a-4-AuthServ-
> HJ97Z4V9TLPH-10278
> 
> Plus additional metrics of 200,400,500s going through the ELBs. More
> relevant to the parent bug, 972065, is that on StackDriver we can set
> threshold alerting. So if latency goes about 500ms for 10+ seconds, we can
> be alerted that there may be an issue.

We can also do alerting w heka, but StackDriver sounds like a good place for it. Also, I noticed that it segments by availability zone. Edwin -- is that good enough for "region"? Benson -- can we add other segmentations in stackdriver, like browser/os/version?

Comment 3

[Benson Wong [:mostlygeek]]

Katie, nope. The stackdriver stuff what is pulled out of Cloudwatch, so basically a single number. If we want to segmentation we would need to use kibana/heka. 

I would like to rescind my use of "same functionality" and say "some functionality" :)

Comment 4

[Edwin Wong [:edwong]]

With both stackdriver and kibana - i would call this fixed.  Thanks!