Skip to content

Latest commit

 

History

History
63 lines (43 loc) · 2.39 KB

SOP.md

File metadata and controls

63 lines (43 loc) · 2.39 KB

Standard operating procedures for the Analytics platform

Issues in end to end tests

Message "Note: integration tests are running localy via docker-compose"

  • it means that some environment variable(s) are not set up properly
  • make sure that the F8A_API_URL is set
  • make sure that the F8A_JOB_API_URL is set

Message "Note: integration tests are running against existing deployment"

  • all environment variables, including F8A_API_URL are set and e2e tests check the remote APIs

Message "Note: {e} environment variable is specified, but tests are still run locally"

  • some of environment variables F8A_API_URL, F8A_JOB_API_URL or F8A_GEMINI_API_URL is setup, but not all
  • it might mean that you need to run end to end tests against remote APIs, but in reality tests are run locally
  • make sure that the F8A_API_URL is set
  • make sure that the F8A_JOB_API_URL is set

Message "Warning: the {name} environment variable is not set. Most tests that require authorization will probably fail"

  • make sure the RECOMMENDER_REFRESH_TOKEN or RECOMMENDER_API_TOKEN is set (but no both)
  • make sure the JOB_API_TOKEN is set in case you need to run jobs API tests (deprecated right now)

Issues on CI

CI jobs failure with message "The requested operation failed as no inventory is available"

Full message looks like this:

+ true
++ cico node get -f value -c ip_address -c comment
The requested operation failed as no inventory is available.
+ cico_output=
Build step 'Execute shell' marked build as failure
Setting status of afca674a4bf1d03ba432d62f108541130905954c to FAILURE with url https://ci.centos.org/job/devtools-e2e-fabric8-analytics/1348/ and message: 'Build finished. '
Using context: ci.centos.org PR build (fabric8-analytics)
Finished: FAILURE
  • this is temporary issue caused by insufficient number of nodes on CI
  • rerun the job after some time by adding [test] message to the pull request

CI jobs failure with messages "cico node get' failed, trying again in Ns (x/y)"

Typical error message is repeated several times:

'cico node get' failed, trying again in 60s (1/15)
+ n=2
+ sleep 60
+ true
++ cico node get -f value -c ip_address -c comment
The requested operation failed as no inventory is available.
  • this is temporary issue caused by insufficient number of nodes on CI
  • rerun the job after some time by adding [test] message to the pull request