Che: Build of E2E testing jobs had failed on Eclipse Che CI because of existed github identity provider

[dmytro-ndp]

Description

We encountered a number of builds failure on Eclipse Che CI because of the same error:
https://ci.codenvycorp.com/view/pr-builds/job/che-pullrequests-test-ocp/293/consoleText
https://ci.codenvycorp.com/view/pr-builds/job/che-pullrequests-test-ocp/294/consoleText
https://ci.codenvycorp.com/view/qa/job/che-integration-tests-multiuser-master-ocp/229/consoleText

[INFO]: Che successfully deployed and is available at http://che-eclipse-che.172.30.3.252.nip.io 
Logging into http://localhost:8080/auth as user admin of realm master
Identity Provider github already exists
Build step 'Execute shell' marked build as failure

[dmytro-ndp]

List of identity providers is empty at the moment:

Check if there is something remained after run...
Logging into http://localhost:8080/auth as user admin of realm master
[ ]

[dmytro-ndp]

Actually, _Github Identity Provider_ remains after destroy and reinstall Eclipse Che on OCP:
https://ci.codenvycorp.com/job/che-pullrequests-test-ocp/300/consoleText

[INFO]: Che successfully deployed and is available at http://che-eclipse-che.172.30.3.246.nip.io 
Check if there is something remained after run...
Logging into http://localhost:8080/auth as user admin of realm master
[ {
  "alias" : "github",
  "internalId" : "73eb2174-3e60-486e-8c1d-8540d7ab0075",
  "providerId" : "github",
  "enabled" : true,
  "updateProfileFirstLoginMode" : "on",
  "trustEmail" : false,
  "storeToken" : true,
  "addReadTokenRoleOnCreate" : true,
  "authenticateByDefault" : false,
  "linkOnly" : false,
  "firstBrokerLoginFlowAlias" : "first broker login",
  "config" : {
    "clientSecret" : "**********",
    "clientId" : "***********",
    "defaultScope" : "repo,user,write:public_key",
    "useJwksUrl" : "true"
  }
} ]

@riuvshin, @eivantsov: could you, please, take a look at this?

[riuvshin]

so, I've added following:

docker run --rm -v /var/lib/origin/:/to_remove alpine sh -c "rm -rf /to_remove/*" || true

which will wipe out all the data that oc cluster up generates on the CI slaves... with this we should never see that issue again. So let's wait for a few days to see if the problem is gone.

[riuvshin]

I do not see that issue anymore on latest builds 308-314
https://ci.codenvycorp.com/view/pr-builds/job/che-pullrequests-test-ocp/

So I consider it is fixed. Feel free to reopen if that will appear again.