Calculate uptime/downtime per day

[Maria.Garcia]

How to calculate the uptime/downtime based on the transactions response?

If the interface is consider as down when 5 consecutive requests not replied within a total of 30 seconds...how to create a rule with this condition?

The time is clear but the number of attempts? How to indicate that value on a condition rule please?

Thanks 

[Eric.Miller]

If I am understanding your question correctly, then you're trying to setup a health rule that triggers if your agent doesn't report in 5 times in a row over a 30 second period. Since the agents only report in every minute, I don't think this will be possible unless AppD adds a "ping" feature which allows users to set their own intervals (a feature I'd love to have too for better Server HealthCheck Monitoring). Your best bet may be to set your health rule to look at the past 1 minute's worth of data and report 1 minute after that if the Agent Availability % drops below 1 or has No data. This won't give you a 30 second response time, but a 120 second one, so you'll have to decide if that extra 90 seconds is acceptable or not.

[Maria.Garcia]

Thanks Eric,

Sorry let me explain better.

We have 3 business transactions that need to be executed in order to complete a login.

The requiremente is:

- if there are 5 attempts of failed login, then there is an application downtime

- an attempt is considered as failed if the business transaction spends more than 30seconds

I hope this clarify the issue...

Thanks

[Eric.Miller]

Ahh that's a bit more complicated. I'd have to think about the first one, but off the top of my head you may be able to open your Application in the Controller GUI, then go to "Configuration" on the left-hand menu then click on "Instrumentation" then "Error Detection" and maybe set some sort of Error using a redirect or an HTTP Code (Maybe 405?), but I'm not sure how your app is designed or your "login" call works. (Error detection documentation can be found here: https://docs.appdynamics.com/display/PRO45/Error+Detection). I'm not sure though I'll have to play with it to see.

As for your second issue, you can do that fairly easier, by creating a Health Rule for that application, using the "Business Transaction Performance (load, response time, slow calls, etc)" Health Rule Type, then select the exact Business Transaction you want, and create a "Critical Condition" using a metric value of > 30 for response time. Then just create a Policy and Action to trigger that rule. Also, depending on how your application is setup, you may be able to do a 2for1 by adding multiple conditions in the Health Rule set to use an "Any" condition to trigger off your 5 failed login attempts to, but it'll probably use an "errors per minute" metric instead.

[Anand.Yadav]

We have 3 business transactions that need to be executed in order to complete a login.

The requiremente is:

- if there are 5 attempts of failed login, then there is an application downtime

- an attempt is considered as failed if the business transaction spends more than 30seconds

I hope this clarify the issue...

Answer: I am assuming that you need it from real user monitoring not from synthetic (it can be easily available from synthetic).

you can define threshold at application level for slow OR very slow transactions 30 secs.

Now group them all three transaction as "Login" , now create a dashboard for it. Select widget "metric" then select particular application as datasource then select metric category "custom"

Now select metric expression from Select a Metric field .

Now select Two metrics as mentioned below:
Login Business transaction group : V2 slow calls metric - Value  &  V1 Average Response Time (ms) - Count

Use below formula , it will give you Availability % for Login BT group :
 100 - (100*({v2}/({v1}+{v2})))

Thanks

[Maria.Garcia]

Thanks Anand,

We cannot use real user monitoring, it is not available for us.

Regarding the formule, sorry could you please kindly clarify it?

 V2 slow calls metric - Value 

 V1 Average Response Time (ms) - Count

[Maria.Garcia]

Hi,

Sorry but I do not undersatnd why  V2 (slow calls metric - Value)  is added to   V1 (Average Response Time (ms) - Count).

You would be adding the number of slow calls to the number of times the agent collected the metric. But those slow calls should be already included in the V1 count... am I right?

Thanks 

[Anand.Yadav]

Correct , my bad.

thanks for correcting it :)

[Maria.Garcia]

Which should be then the right expression to be used?

 100 - (100*({v2}/({v1}))

[Maria.Garcia]

Hi,

Please kindly confirm how to proceed.

Thanks