docker executor的gitlab runner队列不工作

我有：

CentOS 6最终版本
GitLab 11.10.4
带有android executor的Gitlab runner 11.10.1

我的部分

config.toml

：

concurrent = 1
check_interval = 0

[session_server]
  session_timeout = 1800

[[runners]]
  name = "android-1"

image: runmymind/docker-android-sdk:latest

variables:
  GRADLE_USER_HOME: $CI_PROJECT_DIR/.gradle

cache:
  key: $CI_PROJECT_NAME
  paths:
    - .gradle/wrapper/
    - .gradle/caches/

before_script:
  - echo $JAVA_HOME
  - echo $ANDROID_HOME
  - echo $PATH
  - chmod +x ./gradlew

stages:
  - check

checkStage:
  stage: check
  script:
    - set +e
    - du -sh $CI_PROJECT_DIR/.gradle/wrapper
    - du -sh $CI_PROJECT_DIR/.gradle/caches
    - set -e
    - ./gradlew check
  artifacts:
    when: on_failure
    paths:
    - build_results

如何查看

concurrent=1

让我们看看我的

.gitlab ci.yml

：

concurrent = 1
check_interval = 0

[session_server]
  session_timeout = 1800

[[runners]]
  name = "android-1"

image: runmymind/docker-android-sdk:latest

variables:
  GRADLE_USER_HOME: $CI_PROJECT_DIR/.gradle

cache:
  key: $CI_PROJECT_NAME
  paths:
    - .gradle/wrapper/
    - .gradle/caches/

before_script:
  - echo $JAVA_HOME
  - echo $ANDROID_HOME
  - echo $PATH
  - chmod +x ./gradlew

stages:
  - check

checkStage:
  stage: check
  script:
    - set +e
    - du -sh $CI_PROJECT_DIR/.gradle/wrapper
    - du -sh $CI_PROJECT_DIR/.gradle/caches
    - set -e
    - ./gradlew check
  artifacts:
    when: on_failure
    paths:
    - build_results

它起作用了。但当我推进几个分支时，我有： 在一份工作中：

Running with gitlab-runner 11.10.1 (1f513601)
  on android-1 cf4dbddf
Using Docker executor with image runmymind/docker-android-sdk:latest ...
Using locally found image version due to if-not-present pull policy
Using docker image sha256:05edf996f2cf3a0be9c332ca88e4ffd1698b7d59b1c5318b48491033e062e639 for runmymind/docker-android-sdk:latest ...
Running on runner-cf4dbddf-project-918-concurrent-0 via git-ci-runner...
WARNING: Possibly zombie container runner-cf4dbddf-project-918-concurrent-0-predefined-1 is disconnected from network bridge
Reinitialized existing Git repository in /builds/android/MYPROJECT/.git/
Fetching changes...
Checking out 21ff9883 as branch_for_test_it_1...
Removing .gradle/
....
OK

在我的另一份工作中：

Running with gitlab-runner 11.10.1 (1f513601)
  on android-runner-1 cf4dbddf
Using Docker executor with image runmymind/docker-android-sdk:latest ...
Using locally found image version due to if-not-present pull policy
Using docker image sha256:05edf996f2cf3a0be9c332ca88e4ffd1698b7d59b1c5318b48491033e062e639 for runmymind/docker-android-sdk:latest ...
Running on runner-cf4dbddf-project-918-concurrent-0 via git-ci-runner...
Reinitialized existing Git repository in /builds/android/MYPROJECT/.git/
Fetching changes...
Checking out 21ff9883 as branch_for_test_it_2...
ERROR: Job failed (system failure): Error: No such container: 9721b3efbc71ababc12da8606307bd36c56b08e6500605a3ddbc9e42855498d6 (executor_docker.go:976:0s)

我的理解是，每个项目使用一个容器（不适用于分支）。 我的理解是，我们有两种方法来解决这个问题：

1） 了解如何为每个分支创建新容器
2） 了解如何为docker executor创建gitlab runner队列

我想问你，我们如何实现其中一种方法？至于我，最好使用第二种方法，因为在第一种方法中，我们将使用大量的计算机资源。但是没关系。请告诉我，我们如何解决我上面写的这些变体