Reports?

Hi there,

I've found some time to take a look at Gatling again and am working
through getting started.
I first ran the provided BasicExampleSimulation and I have some
questions about the reports.

1) What is the difference between "Global - Requests /sec" and "Global
- Transactions/sec".
    For the basic example the charts are subtly different, but I'm not
sure how and the wiki doesn't seem to mention transactions/sec

2) For the 'Latency' report under the details tab, the latency is
always being reported as 0ms. Huh? How is that possible? Could it be a
bug?

3) It would be great to have the details under 'Statistics' available
at the 'Global' level. (ie aggregated over all request types). Is this
possible?

Thanks for any help.

We are excited about maybe using Gatling - it looks like a great tool :slight_smile:

2012/5/4 Perryn Fowler <pezlists@gmail.com>

Hi there,

I’ve found some time to take a look at Gatling again and am working
through getting started.
I first ran the provided BasicExampleSimulation and I have some
questions about the reports.

  1. What is the difference between “Global - Requests /sec” and "Global
  • Transactions/sec".
    For the basic example the charts are subtly different, but I’m not
    sure how and the wiki doesn’t seem to mention transactions/sec

“Transactions/sec” is the number of response received per sec. I can show you some kinds of bottlenecks on your app.
I’ll fix the wiki today:
https://github.com/excilys/gatling/issues/500

  1. For the ‘Latency’ report under the details tab, the latency is
    always being reported as 0ms. Huh? How is that possible? Could it be a
    bug?

Indeed a bug:
https://github.com/excilys/gatling/issues/499

  1. It would be great to have the details under ‘Statistics’ available
    at the ‘Global’ level. (ie aggregated over all request types). Is this
    possible?

Yep, I added the feature request:
https://github.com/excilys/gatling/issues/501

Thanks for any help.

499 is already fixed and 500 will be shortly. Expect the 1.1.5 release in a few hours.
501 will have to wait for the next one (we usually release every ten days or so).

We are excited about maybe using Gatling - it looks like a great tool :slight_smile:

Thanks!

“Transactions/sec” is the number of response received per sec. I can show you some kinds of bottlenecks on your app.

I’ll fix the wiki today:
https://github.com/excilys/gatling/issues/500

Hi Stephane,

Could you please clear this up a bit more:

  1. Could you please explain failed / successfull requests and failed / successfull transactions? For the transaction(response) I assume the successful may be 200 and for failed - 404 response but what about requests?

  2. What is “All sessions” stand for? Suppose we have a test in which every user does a scenario that consists of three requests. Those are made subsequently with a random pause within one second. So in any particular second of the simulation what would the “All session” parameter present?

Thanks!

Regards,
Eugene

Hi,

1. Could you please explain failed / successfull requests and failed /
successfull transactions? For the transaction(response) I assume the
successful may be 200 and for failed - 404 response but what about requests?

Request vs response charts are about counting the number of requests sent
per second, or the number of responses received per second.
Status is the one of the (request/response) pair, so it's the same for both.
Failure is not only about response status code, for example:

   - what if connection couldn't be established? The request couldn't even
   be sent and that's a failure.
   - if you set up a check on the body because you expect some content and
   it's not here? Check failures are accounted too.
   - what if you do expect another status code? you can override default
   status check.

2. What is "All sessions" stand for? Suppose we have a test in which every
user does a scenario that consists of three requests. Those are made
subsequently with a random pause within one second. So in any particular
second of the simulation what would the "All session" parameter present?

All sessions represent the number of alive users at a given time: started,
but not done.

See documentation here: GitHub - gatling/gatling: Modern Load Testing as Code

Right … so sessions = number of running users, whereas the actual number of sessions on the server side (fully logged in users) may be something else entirely …

You can have both matching if the last step of your scenario is logging out. Otherwise, server session will linger until idle time out.

Hi,
Cloud you please explain detail of latency time in bad network situation? Dose latency mean network consume time?
We want to calculate time without network consume time to fix the real performance issue.

Thanks,

https://github.com/excilys/gatling/wiki/Reports#latency

Latency means the time your server starts responding to a request, from the client POV.
Of course it includes network, Gatling is a client. If you want the server side times, you have to measure there with the proper monitoring tools suited for your technology.