Move the download links to the top and make sticky

[ralfr]

We've created a test that adds a random parameter to the GET URI to avoid caching. This correctly causes Locust to treat every single request as different (exactly what we intend to do).

Unfortunately, this makes the list of requests very very very loooooooong. In order to download the final test results, we'd have to scroll for hours.

Suggestion: Move the download links to the sticky nav bar at the top.

[sanga]

Making those links sticky would probably be nice, but you can solve your problem a different way. locust requests take can take a 'name' parameter with will bin all of those requests in a single bin with that name, irrespective of the url. In that way all of your GET requests will generate only 1 list entry

[heyman]

Yep, I recommend Tim's solution. Otherwise you will also run into memory problems after some time, since locust tracks a bunch of statistics for each URL entry.

Feeling a bit bad since I was the one who gave you the code snippet for adding that random URL parameter, and I completely forgot to add the name argument. Here's an updated version:

self.client.get("/some/path?rand=%i" % random.randint(0, 1000000), name="/some/path?rand=[random-number]")

[ralfr]

No worries, glad there are folks here who are willing to help. One more thing: Is there any way to set a different host on a per task level?

I understand that via self.client.auth the authentication can be changed on a per task level and it works like a charm.

Unfortunately, self.client.host does not work an giving self.client.get() an absolute URI, including protocol and host doesn't either.

Any idea?

[heyman]

I you give a full URL (including hostname) to client.get(), the request will be made to that host. However, this won't be visible in the statistics unless you use the name argument to make it so.

Like this:

self.client.get("http://other.host.com/some/path", name="http://other.host.com/some/path")

[ralfr]

I think all of this would be worth to be added to the docs at locust.io :-)

[heyman]

I agree :). I've added a section about grouping requests using the name argument now: http://docs.locust.io/en/latest/writing-a-locustfile.html#grouping-requests-to-urls-with-dynamic-parameters