Possible bug: ERR_APP_ALREADY_DEPLOYED


#1

Hi, while testing eventing functions I managed to get the cluster in a weird state.

My test does:

  1. Create a function (let’s call it fn).
  2. While it’s being bootstrapped, set settings for it to become undeployed.
  3. Repeatedly call delete endpoint until fn gets deleted.
  4. Create another function fn with different content.
  5. While it’s being bootstrapped, set settings for it to become undeployed.
  6. Repeatedly call delete endpoint until fn gets deleted.

After this point, I tried to run again the same test, and immediately got an error ERR_APP_ALREADY_DEPLOYED, even if in the UI no functions are visible, and the REST API returns an empty list for current functions. The cluster is now in this weird state.

Response:

{
  "name": "ERR_APP_ALREADY_DEPLOYED",
  "code": 20,
  "description": "Function already deployed",
  "attributes": null,
  "runtime_info": {
    "code": 20,
    "info": "Function: db_eventing_integration_test_testFn_functionLifecycle_sameName another function with same name is already deployed, skipping save request"
  }
}

Here are the logs at the current state: https://s3.amazonaws.com/cb-customers/Alberto+Marchetti/collectinfo-2018-09-04T175138-ns_1%40127.0.0.1.zip

Thanks,
Alberto


#2

Hi Alberto,

I suspect some problem was there in beta build. In current stable builds, we’re repeatedly test somewhat similar scenario on every commit - test case. So suspect similar problem won’t occur in latest builds. I anyways have tried the test in loop, problem doesn’t occur of newer builds.

Thanks,
Abhishek