triggering against web scenario/step

[Yello]

Hi,
Ealier today I managed to satisfy myself that the "Required" field of the web step configuration actually does match text from a given page. I did this by changing the text to something random and noted an error reported on the web monitoring screen.

However, that raises an interesting question. The web scenario I've configured has created three items:

- Response time
- Download speed
- Response code

If I create a trigger against the response code item then it always succeeds because the return code is always 200 regardless of whether the search string is found or not. So, can someone explain how I can trigger off a web scenario when the "required" string is not matched?


Regards,
David

[Rudd]

Hi David,

When creating the trigger look for the item "Failed step of scenario". That will trigger when the "Required" value does not match, or any other test of the scenario fails.

The resulting trigger should look something like this:

{www.site.com:web.test.fail[Web Application].last(0)}#0

[Yello]

Thank you. I'll try that out a little later.

Can I also ask about the rationale behind web steps? Are these simply a mechanism for group similar actions, in this case string searches? Or are they intended for group serially related actions that should be performed one after the other?

I am assuming it's the latter. Is that right?


Regards,
David

[Rudd]

That is correct. It allows you to create a step by step scenario

[mushero]

Hi David,

When creating the trigger look for the item "Failed step of scenario". That will trigger when the "Required" value does not match, or any other test of the scenario fails.

Great, but how to tell the difference between a timeout, text mismatch, and return code mismatch - we want to trigger separately on each situation.

[Yello]

Hi,
Yes, I use separate triggers for each. And then overlay those with a master trigger so that only 1 failure will generate a callout alarm. Otherise you can suffer "callout storms" where callout would suffice, when several steps of several scenarios fail.


Regards,
David

[mushero]

Would love to see your trigger examples as I spent quite a bit of time but couldn't figure out how to separate them, in part because the data items are all zero if there is any failure in the step, i.e. return code & response time are zero so I can't tell the difference between the error types. A bit thanks if you can help.

[Yello]

Would love to see your trigger examples as I spent quite a bit of time but couldn't figure out how to separate them, in part because the data items are all zero if there is any failure in the step, i.e. return code & response time are zero so I can't tell the difference between the error types. A bit thanks if you can help.

Show me what you've done so far.

[mushero]

Deleted them all now as I gave up.

We are interested in at least three situations:
- Timeout, bad host name, etc. - all failure to connect
- Bad response code, such as 5xx
- Bad text match

Seems we have a few pieces of data:

Response time
Response code
Failure step

Our standard trigger is:
{srv-xxx-web1:web.test.fail[www.xxx.com].last(0)}>0

This triggers on any failure.

My original thought was to combine them, such as this logic:
- test.fail() and Response time > 30 = response time problem
- test.fail() and Response code <> 200 = response code problem
- test.fail() and NOT (Response time > 30 OR Response Code <> 200) = timeout

But if test.fail() happens the response time and code are zero, so useless, it seems.

I saw some note on testing the step of failure but not sure how that helps.

Would also love access to the text of failure, shown in the Status column of web monitoring as that has info on real failure such as mismatch, DNS lookup, etc.

[yong1515yong]

This posts did help a lots!
Thanks!

[Yello]

Deleted them all now as I gave up.

We are interested in at least three situations:
- Timeout, bad host name, etc. - all failure to connect

These are all picked up by web.test.fail

- Bad response code, such as 5xx

Picked up by web.test.rspcode and web.test.fail (if I recall)

- Bad text match

Picked up by web.test.fail.

Seems we have a few pieces of data:

Response time
Response code
Failure step

Our standard trigger is:
{srv-xxx-web1:web.test.fail[www.xxx.com].last(0)}>0

This triggers on any failure.

My original thought was to combine them, such as this logic:
- test.fail() and Response time > 30 = response time problem

In my experience this would never resolve to true. Many "fail" scenarios would lead to "too fast to be true" response times.

- test.fail() and Response code <> 200 = response code problem
- test.fail() and NOT (Response time > 30 OR Response Code <> 200) = timeout

It seems like you want to use trigger combinations to define what the problem is. I'm not sure that's practical at the moment.

My triggers look like this for most tests, and work quite well:

1. {svr1:web.test.fail[test1].min(#3)}#0 | {svr1:web.test.rspcode[test1,step1].count(#3,200,"eq")}=0
2. ({TRIGGER.VALUE}=0 & {svr1:web.test.time[test1,step1,resp].min(#3)}>5) | ({TRIGGER.VALUE}=1 & {svr1:web.test.time[test1,step1,resp].max(#3)}>3)

We have several triggers formatted like this. Then I have a master trigger, which i won't print because it is too long which will trigger a callout if any of the underlying triggers fire. That is, if one or more tests fail or responds too slowly then raise a callout.

Would also love access to the text of failure, shown in the Status column of web monitoring as that has info on real failure such as mismatch, DNS lookup, etc.

Yeah, I don't think that's possible. Although you could go direct to the database for the information I imagine. Raise that as a feature request if you think it would be useful.