Add HttpWaitStrategy

[gossi]

Impl for #236

I originally build this for Req. Then saw, there is Tesla. I followed their docs to build the client and make a request - I never used tesla before. I checked the response type from tesla and req and they are pretty equal for the response.

While I tested this with Req, this PR is completely untested - but I thought it's good enough to say hello :)

[jarlah]

looks useful :) should be very easy to test it in a isolated manner outside a container configuration. Just start a container with the exunit macro and then construct and call the strategy ? then another test would be to create a special container, maybe an nginx container, that serves a specific html file or something. However it seems to always take a map in the match function. Is it a requirement as the strategy is now that response must be json ? Should it require that?

[gossi]

I did a test-case, verifying the default index.html file from nginx is up.

Is it a requirement as the strategy is now that response must be json ? Should it require that?

I don't think at all. You are free to pass headers and a matcher to adjust to your setting.

[gossi]

And I hit a timeout on CI. Quite ironic, it is a :socket_closed_remotely error, addressing that is the reason for this PR 😅

But apparently this is a layer lower - any idea for this?

[jarlah]

being so random, i try to run them again. they told me nothing. It could be the engine, testcontainers calling the strategy, but meh .. i will think about it

[gossi]

I think what happens is: Tesla makes a request to a non-yet-existing endpoint and waits the timeout for it to return. What we want is: Use the timeout window to retry until the request passes.

Req has retry on by default (with exponential backoff) - Tesla has a ready-to-use middleware. I think, we want then this config:

• max_retries - default to: 3?
• timeout - default to: 10_000

And then in the request:

• :timeout (middleware) - timeout / max_retries
• no delay
• no exponential backoff
• no jitter

[gossi]

That might have done it. I think the failures are coming from unrelated tests? (I cannot re-run them)

[jarlah]

think there is a direct failure here

** (FunctionClauseError) no function clause matching in Testcontainers.WaitStrategy.Testcontainers.HttpWaitStrategy.validate_response/1
(testcontainers 2.0.0-rc) lib/wait_strategy/http_wait_strategy.ex:98: Testcontainers.WaitStrategy.Testcontainers.HttpWaitStrategy.validate_response({:error, :socket_closed_remotely})
(testcontainers 2.0.0-rc) lib/wait_strategy/http_wait_strategy.ex:86: Testcontainers.WaitStrategy.Testcontainers.HttpWaitStrategy.wait_until_container_is_ready/3
(elixir 1.18.4) lib/enum.ex:2546: Enum."-reduce/3-lists^foldl/2-0-"/3
(testcontainers 2.0.0-rc) lib/testcontainers.ex:374: Testcontainers.create_and_start_container/3
(testcontainers 2.0.0-rc) lib/testcontainers.ex:181: anonymous fn/3 in Testcontainers.handle_call/3
(elixir 1.18.4) lib/task/supervised.ex:101: Task.Supervised.invoke_mfa/2
(elixir 1.18.4) lib/task/supervised.ex:36: Task.Supervised.reply/4
Function: #Function<1.21825989/0 in Testcontainers.handle_call/3>
Args: []

but it could also be wierd corner case, due to timeout of course

[jarlah]

i think the toxiproxy containers test is a bit flaky too .. so some flaky tests happen. its a symptom but i dont have time to dig into everything :) i will rerun one more time

[jarlah]

copilot suggest this

change this

defp validate_response({:ok, response}), do: response

to this

defp validate_response({:ok, response}), do: response
defp validate_response({:error, reason}), do: {:error, reason}

[jarlah]

has to be something with that specific elixir version or something ...

[jarlah]

ill try to rerun all ;)

[gossi]

copilot suggestion was false. Actually, that's the "case falling through" in the with clause then catched already - but always worth a try when out of options 😂

The weird part: This PR fails on tests, that "never" use the code added in here o_O

[jarlah]

Well actually the error disappeared. Its only toxiproxy tests failing. And they failed before from time too. I think you are in the clear. I will look into the failing and irrelevant tests

[jarlah]

Marked the toxiproxy test as flaky and updated this pr