How to use a single OAuth2.0 token for Multiple Virtual Users in a Gatling load test

I need to load test an API that requires an OAuth2.0 token via Gatling (of which I'm a complete novice!) but would like each virtual user to use the same token. I'm retrieving the token ok (I think) and putting it in a variable called 'access' but I keep getting 'no attribute named 'access' is defined' when the test itself starts.

My token retrieval looks like the following(along with httpConf, used below):

 class MySimulation extends Simulation {

 val httpConf = http        
    .baseUrl("https://MyBaseUrl.Com/")
    .acceptHeader("application/json") 
    .doNotTrackHeader("1")
    .acceptLanguageHeader("en-UK,en;q=0.5")
    .acceptEncodingHeader("gzip, deflate")
    .userAgentHeader("Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:16.0) Gecko/20100101 Firefox/16.0")
    .shareConnections

 val header = Map("Content-Type" -> """application/x-www-form-urlencoded""")

 al auth = scenario("Retrieve Token")
 .exec(http("POST OAuth Req")
 .post("https://SomeTokenUrl")
 .formParam("resource", "someresource")
 .formParam("grant_type", "somegranttype")
 .formParam("client_secret", "someclientsecret")
 .formParam("client_id", "someclientid")
 .headers(header).check(status.is(200)).check(jsonPath("$.access_token").find.saveAs("access")))

I then tried setting up the load test as (Note: I did initially put 'Map', rather than the mutable variant, but read somewhere the default was immutable, and wondered if this was why the header couldn't update):

 val headers_10 = scala.collection.mutable.Map("Content-Type" -> "application/json; charset=ISO-8859-1", "Authorization" -> "Bearer ${access}")

 val scn = scenario("MyService Gatling test run")       
           .exec(http("")               
           .post("Myservice/api")
           .headers(headers_10.toMap)                
           .body(StringBody("""{"SomeProperty": "Some Value"}"""))
           .asJson
           .check(status.is(200)))

 setUp(
    auth.inject(constantUsersPerSec(1) during (2 seconds)),
    scn.inject(nothingFor(2 seconds),
    constantUsersPerSec(10) during (10 seconds) 
    ).protocols(httpConf))
    .assertions(global.responseTime.max.lt(500)) 
    .assertions(forAll.failedRequests.percent.lte(1)) 
    .assertions(global.responseTime.mean.lte(100))

The idea was that the token retrieval would complete prior to the load test kicking in and the 'access' variable would then be used by the load test scenario, but it gives the following result:

  ERROR : Failed to build request: No attribute named 'access' is defined

I've reached the end of my tether with it. I'm guessing it might be something to do with scopes, and perhaps the variable doesn't carry over to the load test scenario, but I've seen examples elsewhere that seem to recommend exactly that set up, so I don't know whether these other examples are partially complete or what.

Solution

Sessions are per user and no session data is shared between users. So while you have 1 user running your 'auth' scenario and saving the token, it is two different users that run 'scn' and they don't have access to the session values of the auth user.

It's not recommended practice, but you can solve this by pushing the auth token into a regular scala var and the setting this in the auth scenario and reading it in the main scenario - you just need to be sure that auth always completes before you inject any other users.

var token: String = ""

then in the auth scenario, have a step at the end such as

.exec(session => {
    token = session("access").as[String]
    session
})

then at the start of the scn scenario have a step to set the session variable

.exec(session.set("access", token))

I've used this pattern in the past and it works, but I'm sure there are nicer ways to do it