[MERGE-AFTER-PR-5366] Multi-container workspace Support #5110

davidfestal · 2017-05-16T21:03:28Z

What does this PR do?

Fixes the support of multi-container (multi-machine) Che workspaces inside Openshift.

In the current design of the OpenShift connector, each Che machine create is transformed into an OpenShift Pod with a single container. (and not a Pod with many containers).

However, the multi-machine use-case wasn't fully tested and NPEs used to prevent the setup of non-development additional machines. This is now fixed.

Additionally, since machines live in distinct pods, the docker compose internal network feature cannot be used to allow machines communicating with their compose service names.
To fix this, the OpenShift service name, for all non-development Che machines, will be created with the name of the related compose service. This way we can leverage the OpenShift internal Dns server that allows all pods in an OS namespace to communicate through their OS service name.

What issues does this PR fix or reference?

https://issues.jboss.org/browse/CHE-175
https://issues.jboss.org/browse/CHE-258
https://issues.jboss.org/browse/CHE-259

codenvy-ci · 2017-05-16T21:04:02Z

Can one of the admins verify this patch?

davidfestal · 2017-05-16T21:05:42Z

@l0rd, @gorkem, could you review this PR please ?

l0rd

If you have time you can extract the for loops outside the if to make code more easy to read. But otherwise LGTM

l0rd · 2017-05-22T13:45:12Z

...c/main/java/org/eclipse/che/plugin/openshift/client/kubernetes/KubernetesLabelConverter.java

-                    LOG.error(
-                            "Could not convert label {} into Kubernetes annotation: labels must be alphanumeric with ':' and '/'",
-                            label.toString());
+        if (labels != null) {


Avoiding to nest the for loop inside the if would make it more readable:

if (labels == null) { returns names; } for (Map.Entry<String, String> label : labels.entrySet()) { ... }

l0rd · 2017-05-22T13:45:45Z

...c/main/java/org/eclipse/che/plugin/openshift/client/kubernetes/KubernetesLabelConverter.java

-            }
-            key = keyMatcher.group(1);
-            value = valueMatcher.group(1);
+        if (names != null) {


Same comment here than above.

l0rd · 2017-05-22T13:48:35Z

cc @sunix

sunix · 2017-05-22T14:07:39Z

ide/che-core-ide-stacks/src/main/resources/stacks.json

@@ -1,5 +1,74 @@
 [
  {
+    "id": "java-centos-mysql",


is this new stacks supposed to go upstream ?

I don't know. @l0rd can tell.

@davidfestal @sunix it makes sense to me to merge this stack to eclipse/che master

davidfestal · 2017-05-23T16:58:31Z

@l0rd I committed e66d31d as you suggested. Do you think you could you merge ?

sunix · 2017-05-24T09:19:57Z

ide/che-core-ide-stacks/src/main/resources/stacks.json

+          "machines": {
+            "dev-machine": {
+              "agents": [
+                "org.eclipse.che.terminal", "org.eclipse.che.ws-agent", "com.redhat.bayesian.lsp"


if this is supposed to go to master, it shouldn't include com.redhat.bayesian.lsp agent

That's a good point @sunix. I think @davidfestal did it for consistency since other stacks use the same agent here. But I understand that's an issue for you because you are in the middle of a rebasing. I think we have 2 choices:

@davidfestal remove com.redhat.bayesian.lsp and we merge it right now

@davidfestal change the PR to be merge on master branch instead of openshift-connector

@sunix are these 2 options ok for you?
@davidfestal which of these 2 options is better for you?

When I say change the PR to be merged on master I actually mean to split this PR in two:

one PR for master with the modifications of stacks.json

one PR for openshift-connector with modifications on the OpenShiftConnector

davidfestal · 2017-05-24T11:30:34Z

@l0rd

remove com.redhat.bayesian.lsp and we merge it right now

Let me choose this option. Then we'll be able to merge on openshift-connector branch right now, right ?

davidfestal · 2017-05-24T11:53:41Z

@l0rd @sunix I changed the stacks file to remove the default activation of Bayesian. Is it OK to merge it ?

sunix · 2017-05-24T16:20:25Z

The second problem is that in the branch openshift-connector-rebased and openshift-connector-wip we did reverted all changes on stack.json, so if you could cherry-pick and fix conflicts + send PR (to openshift-connector-rebased) it would work for me. + we need to plan a task so the changes are applied in rh-che stacks.json as well

davidfestal · 2017-05-24T17:36:34Z

@sunix. Not sure I fully understand what you're proposing. But let me remove the stacks.json from this PR and simply add the stacks.json change in a PR against master, since @l0rd said the newly-created java-mysql-centos stack was expected to finally be merged to eclipse/che master. This shouldn't bring any conflict like that, right ? Of course we'll have to reintegrate this new stack into the rh-che stacks.json file.

@l0rd wdyt ?

davidfestal · 2017-05-29T12:27:29Z

@sunix I removed the stacks.json from this PR and created a new PR for it (#5227).
@l0rd I added the changes that also fix https://issues.jboss.org/browse/CHE-258 and https://issues.jboss.org/browse/CHE-259

Is it possible to merge this ?

l0rd · 2017-05-30T16:30:34Z

@davidfestal can you please rebase this on top of branche openshift-connector-rebased

davidfestal · 2017-05-31T11:21:10Z

@l0rd @sunix I rebased to the openshift-connector-rebased branch

However this should be rebuilt and tested again when we'll be ready to merge to openshift-connector-branch or to master.

In the context of https://issues.jboss.org/browse/CHE-175 Signed-off-by: David Festal <dfestal@redhat.com>

This fixes https://issues.jboss.org/browse/CHE-259 and https://issues.jboss.org/browse/CHE-258 Signed-off-by: David Festal <dfestal@redhat.com>

davidfestal · 2017-06-15T10:45:07Z

@sunix could you review this PR ? it should be able to be merged without any problem / conflicts now.

benoitf · 2017-06-16T08:07:12Z

maybe in the description, it could say that each Eclipse Che machine is transformed into an OpenShift Pod with a single container. (and not a Pod with many containers)

davidfestal · 2017-06-16T08:08:39Z

maybe in the description, it could say that each Eclipse Che machine is transformed into an OpenShift Pod with a single container. (and not a Pod with many containers)

Sure let me add a more detailed explanation

* Fix 2 NPE that prevented using *non-dev* additional machines In the context of https://issues.jboss.org/browse/CHE-175 Signed-off-by: David Festal <dfestal@redhat.com> * Name openshift resources based on the machine name for non-dev machines This fixes https://issues.jboss.org/browse/CHE-259 and https://issues.jboss.org/browse/CHE-258 Signed-off-by: David Festal <dfestal@redhat.com>

@fbenoit

* CHE-4141 - Use Persistent Volumes Claims when creating workspaces Signed-off-by: Snjezana Peco <snjezana.peco@redhat.com> * Implement getContainerLogs method in OpenShiftConnector Signed-off-by: Snjezana Peco <snjezana.peco@redhat.com> * Implement createExec() and startExec() in OpenShiftConnector Add implementations of createExec() and startExec(). Since OpenShift does not separate the create and start steps, a holder class KubernetesExecHolder is necessary, to pass information between the call to createExec() (which just saves relevant information) and startExec(). Additionally, adds KubernetesOutputAdapter, which parses the output from OpenShift into LogMessages that can be handled by Che's MessageProcessor<LogMessage> class. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Add implementation of getEvents() to avoid busy wait Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Update Dockerfile to avoid permissions issues Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Che server and workpaces exposed on the same single TCP port (#4351) Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Disabling usage of user account service in openshift-connector Signed-off-by: Sun Seng David Tan <sutan@redhat.com> * Update Docker Compose tests to fix test failure Updating to Jackson 2.7.7 causes tests in the docker compose plugin to fail. This is due to the fact that the tests expect empty values in dictionaries to be parsed as the empty string, whereas jackson 2.7.7 parses them as null (as specified by the yaml spec). Modifies the affected tests to explicitly use an empty string (i.e. "") instead of an empty value. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Find an alternative to subPath in volumeMount Signed-off-by: Snjezana Peco <snjezana.peco@redhat.com> * Setting rwx permissions for all on /data/ in case it's not mounted Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Add support for resource limits when running on Openshift Add resource limits to workspace Pods when running on OpenShift. The memory limit is normally obtained from the API request to create the workspace, however it can be overridden via the property `che.openshift.workspace.memory.override`. The cpu limit used is determined by the property `che.openshift.workspace.cpu.limit`. In both cases, the value of the property is passed directly to OpenShift, so any valid quantity is acceptable (e.g. 150Mi, 1Gi, 1024, etc). Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Fix dockerImageConfig is null (since v1.5 of OpenShift API) Signed-off-by: Sun Seng David Tan <sutan@redhat.com> * Add Nullable annotation to che.docker.ip.external The property che.docker.ip.external can be null, but OpenShiftConnector does not include the annotation. This prevents Che from initialising if e.g. running on docker without the property set. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * CHE-158 Adding TLS support for Workspace routes Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Adding property to set requests for RAM Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * CHE-158 Using '-' instead of '.' for generating OpenShift route Urls Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Fixing tests after changing Url generation logic Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Redirect insecure HTTP requests to TLS endpoint Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * CHE-180: Creating and closing OpenShiftClient in every method of OpenshiftConnector Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Update route naming to make it work on OSO Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Rework PVC management on OpenShift - Change how subdirectories are created in pods to use a short, terminating job instead of a full deployment - Add OpenShiftWorkspaceFilesCleaner class to properly notice workspace deleted events - Add helper class to manage job pods. For creation, some effort is made to avoid attempting to create workspaces unnecessarily, but only exists in-memory - Workspace deletions are batched together so that removing workspaces directories can be done when server is idled, avoiding unnecessary PVC mounts - Add two new properties: che.openshift.jobs.image and che.openshift.jobs.memorylimit, which are used by OpenShiftPvcHelper to set up pods Current issues: - Since workspace directories are not deleted immediately, attempting to re-create a workspace with the same name will result in the previous instance's project to already be there. This should have a minor impact. - Memory for which workspace dirs have been created is not persisted, resulting in potentially unnecessary jobs - Openshift workspace files cleaner is included by overwriting binding in WsMasterModule instead of using a provider. This could be better, but OpenShift integration may be reaching a point where a custom module is a better solution. Signed-off-by: Angel Misevski <amisevsk@redhat.com> Signed-off-by: Sun Seng David Tan <sutan@redhat.com> * Delete ReplicaSets explicitly when shutting down a workspace Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Fix OpenShiftConnectorTest Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Fix route server names if unknown should start with server-. https://issues.jboss.org/browse/CHE-230 Signed-off-by: Sun Seng David Tan <sutan@redhat.com> * Add property to control manual workspace dir creation in OpenShift Add property 'che.openshift.precreate.workspace.dirs'. If property is true, OpenShiftConnector will run a pod before launching workspaces to create a subpath in the workspace's persistent volume with correct permissions. If the property is false, this step is skipped. This is necessary as in older versions of OpenShift/Kubernetes, subpaths created as part of a volume mount are created with root permissions, and so cannot be modified by workspace pods. More recent versions fix this, creating subpath volumes with correct permissions, making the step above unnecessary. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * CHE-102 - Idle detection of che-server and workspaces Signed-off-by: Snjezana Peco <snjezana.peco@redhat.com> * Add and modify tests for OpenShift helper classes Add tests for the untested classes in openshift.client.kuberentes, and update existing tests where necessary. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Recent changes required access to `/` which is impossible under OS Signed-off-by: David Festal <dfestal@redhat.com> * adapt che-server entrypoint.sh to environments without write permissions in '/' (#5344) * adapt che-server entrypoint.sh to environments without write permissions in '/' * CHE-280: Adding container's state info to the 'inspectContainer' API Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Factorize code of `ServerEvaluationStrategy` classes, to use the Custom strategy as the basis of other strategies (#5366) * Pull-up the local docker port management (use exposed ports) Signed-off-by: David Festal <dfestal@redhat.com> * Make all the strategies extend `CustomEvaluationStrategy` Signed-off-by: David Festal <dfestal@redhat.com> * Add a `workspaceIdWithoutPrefix` macro and use it for `single-port` This macro is based on the `workspaceId` macro, but without the `workspace` prefix. Signed-off-by: David Festal <dfestal@redhat.com> * Add the `isDevMachine` to allow conditions in the ST template. This is required to allow the `single-port` strategy to have a different url according to the type of machine. (see the work done for CHE-175 : Support multi-container workspaces on OpenShift) Signed-off-by: David Festal <dfestal@redhat.com> * Small fixes after comments from @fbenoit Signed-off-by: David Festal <dfestal@redhat.com> * Fix unnecessary space pointed out by @sunix Signed-off-by: David Festal <dfestal@redhat.com> * Remove unnecessary `else` as suggested by @sunix Signed-off-by: David Festal <dfestal@redhat.com> * Keep the method signatures compatible with the `condenvy` strategy Signed-off-by: David Festal <dfestal@redhat.com> * Align names of parameters of constructors (requested by @garagatyi) Signed-off-by: David Festal <dfestal@redhat.com> * Add a default implementation to avoid breaking the Codenvy build Signed-off-by: David Festal <dfestal@redhat.com> * Also rename the attributes Signed-off-by: David Festal <dfestal@redhat.com> * Use a constant for the `workspace` prefix string Signed-off-by: David Festal <dfestal@redhat.com> * Fix formatting as requested by @sunix Signed-off-by: David Festal <dfestal@redhat.com> * Use a constant for the `isDevMachine` macro name Signed-off-by: David Festal <dfestal@redhat.com> * Add unit tests for `workspaceIdWithoutPrefixè and `isDevMachine` macros Signed-off-by: David Festal <dfestal@redhat.com> * Another requested formatting fix Signed-off-by: David Festal <dfestal@redhat.com> * Make new tests clearer Signed-off-by: David Festal <dfestal@redhat.com> * yet another formatting request Signed-off-by: David Festal <dfestal@redhat.com> * Respect the original order of imports Signed-off-by: David Festal <dfestal@redhat.com> * remove unnecessary `toString()` Signed-off-by: David Festal <dfestal@redhat.com> * use a lowercase `S` in the `server-` prefix Signed-off-by: David Festal <dfestal@redhat.com> * Multi-container workspace Support (#5110) * Fix 2 NPE that prevented using *non-dev* additional machines In the context of https://issues.jboss.org/browse/CHE-175 Signed-off-by: David Festal <dfestal@redhat.com> * Name openshift resources based on the machine name for non-dev machines This fixes https://issues.jboss.org/browse/CHE-259 and https://issues.jboss.org/browse/CHE-258 Signed-off-by: David Festal <dfestal@redhat.com> * Fix failing Traeffik tests... ... by: - adding the new `CHE_IS_DEV_MACHINE` env variable in tests - pulling up all the `CustomServerEvaluationStrategy` features in an abstract `BaseServerEvaluationStrategy` (which all other Che strategies extend) and have the `CustomServerEvaluationStrategy` class simply extend this `BaseServerEvaluationStrategy`. Signed-off-by: David Festal <dfestal@redhat.com> * Fix tests in the LocalDockerEvaluationStrategy... ... by correctly using the boolean attribute to manage the new use-case introduced by @fbenoit in master. Signed-off-by: David Festal <dfestal@redhat.com> * Replace OSIO-specific `single-port` strategy by `docker-local-custom` This fixes redhat-developer/rh-che#113 Signed-off-by: David Festal <dfestal@redhat.com>

@fbenoit

* CHE-4141 - Use Persistent Volumes Claims when creating workspaces Signed-off-by: Snjezana Peco <snjezana.peco@redhat.com> * Implement getContainerLogs method in OpenShiftConnector Signed-off-by: Snjezana Peco <snjezana.peco@redhat.com> * Implement createExec() and startExec() in OpenShiftConnector Add implementations of createExec() and startExec(). Since OpenShift does not separate the create and start steps, a holder class KubernetesExecHolder is necessary, to pass information between the call to createExec() (which just saves relevant information) and startExec(). Additionally, adds KubernetesOutputAdapter, which parses the output from OpenShift into LogMessages that can be handled by Che's MessageProcessor<LogMessage> class. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Add implementation of getEvents() to avoid busy wait Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Update Dockerfile to avoid permissions issues Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Che server and workpaces exposed on the same single TCP port (eclipse-che#4351) Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Disabling usage of user account service in openshift-connector Signed-off-by: Sun Seng David Tan <sutan@redhat.com> * Update Docker Compose tests to fix test failure Updating to Jackson 2.7.7 causes tests in the docker compose plugin to fail. This is due to the fact that the tests expect empty values in dictionaries to be parsed as the empty string, whereas jackson 2.7.7 parses them as null (as specified by the yaml spec). Modifies the affected tests to explicitly use an empty string (i.e. "") instead of an empty value. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Find an alternative to subPath in volumeMount Signed-off-by: Snjezana Peco <snjezana.peco@redhat.com> * Setting rwx permissions for all on /data/ in case it's not mounted Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Add support for resource limits when running on Openshift Add resource limits to workspace Pods when running on OpenShift. The memory limit is normally obtained from the API request to create the workspace, however it can be overridden via the property `che.openshift.workspace.memory.override`. The cpu limit used is determined by the property `che.openshift.workspace.cpu.limit`. In both cases, the value of the property is passed directly to OpenShift, so any valid quantity is acceptable (e.g. 150Mi, 1Gi, 1024, etc). Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Fix dockerImageConfig is null (since v1.5 of OpenShift API) Signed-off-by: Sun Seng David Tan <sutan@redhat.com> * Add Nullable annotation to che.docker.ip.external The property che.docker.ip.external can be null, but OpenShiftConnector does not include the annotation. This prevents Che from initialising if e.g. running on docker without the property set. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * CHE-158 Adding TLS support for Workspace routes Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Adding property to set requests for RAM Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * CHE-158 Using '-' instead of '.' for generating OpenShift route Urls Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Fixing tests after changing Url generation logic Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Redirect insecure HTTP requests to TLS endpoint Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * CHE-180: Creating and closing OpenShiftClient in every method of OpenshiftConnector Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Update route naming to make it work on OSO Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Rework PVC management on OpenShift - Change how subdirectories are created in pods to use a short, terminating job instead of a full deployment - Add OpenShiftWorkspaceFilesCleaner class to properly notice workspace deleted events - Add helper class to manage job pods. For creation, some effort is made to avoid attempting to create workspaces unnecessarily, but only exists in-memory - Workspace deletions are batched together so that removing workspaces directories can be done when server is idled, avoiding unnecessary PVC mounts - Add two new properties: che.openshift.jobs.image and che.openshift.jobs.memorylimit, which are used by OpenShiftPvcHelper to set up pods Current issues: - Since workspace directories are not deleted immediately, attempting to re-create a workspace with the same name will result in the previous instance's project to already be there. This should have a minor impact. - Memory for which workspace dirs have been created is not persisted, resulting in potentially unnecessary jobs - Openshift workspace files cleaner is included by overwriting binding in WsMasterModule instead of using a provider. This could be better, but OpenShift integration may be reaching a point where a custom module is a better solution. Signed-off-by: Angel Misevski <amisevsk@redhat.com> Signed-off-by: Sun Seng David Tan <sutan@redhat.com> * Delete ReplicaSets explicitly when shutting down a workspace Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Fix OpenShiftConnectorTest Signed-off-by: Mario Loriedo <mloriedo@redhat.com> * Fix route server names if unknown should start with server-. https://issues.jboss.org/browse/CHE-230 Signed-off-by: Sun Seng David Tan <sutan@redhat.com> * Add property to control manual workspace dir creation in OpenShift Add property 'che.openshift.precreate.workspace.dirs'. If property is true, OpenShiftConnector will run a pod before launching workspaces to create a subpath in the workspace's persistent volume with correct permissions. If the property is false, this step is skipped. This is necessary as in older versions of OpenShift/Kubernetes, subpaths created as part of a volume mount are created with root permissions, and so cannot be modified by workspace pods. More recent versions fix this, creating subpath volumes with correct permissions, making the step above unnecessary. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * CHE-102 - Idle detection of che-server and workspaces Signed-off-by: Snjezana Peco <snjezana.peco@redhat.com> * Add and modify tests for OpenShift helper classes Add tests for the untested classes in openshift.client.kuberentes, and update existing tests where necessary. Signed-off-by: Angel Misevski <amisevsk@redhat.com> * Recent changes required access to `/` which is impossible under OS Signed-off-by: David Festal <dfestal@redhat.com> * adapt che-server entrypoint.sh to environments without write permissions in '/' (eclipse-che#5344) * adapt che-server entrypoint.sh to environments without write permissions in '/' * CHE-280: Adding container's state info to the 'inspectContainer' API Signed-off-by: Ilya Buziuk <ibuziuk@redhat.com> * Factorize code of `ServerEvaluationStrategy` classes, to use the Custom strategy as the basis of other strategies (eclipse-che#5366) * Pull-up the local docker port management (use exposed ports) Signed-off-by: David Festal <dfestal@redhat.com> * Make all the strategies extend `CustomEvaluationStrategy` Signed-off-by: David Festal <dfestal@redhat.com> * Add a `workspaceIdWithoutPrefix` macro and use it for `single-port` This macro is based on the `workspaceId` macro, but without the `workspace` prefix. Signed-off-by: David Festal <dfestal@redhat.com> * Add the `isDevMachine` to allow conditions in the ST template. This is required to allow the `single-port` strategy to have a different url according to the type of machine. (see the work done for CHE-175 : Support multi-container workspaces on OpenShift) Signed-off-by: David Festal <dfestal@redhat.com> * Small fixes after comments from @fbenoit Signed-off-by: David Festal <dfestal@redhat.com> * Fix unnecessary space pointed out by @sunix Signed-off-by: David Festal <dfestal@redhat.com> * Remove unnecessary `else` as suggested by @sunix Signed-off-by: David Festal <dfestal@redhat.com> * Keep the method signatures compatible with the `condenvy` strategy Signed-off-by: David Festal <dfestal@redhat.com> * Align names of parameters of constructors (requested by @garagatyi) Signed-off-by: David Festal <dfestal@redhat.com> * Add a default implementation to avoid breaking the Codenvy build Signed-off-by: David Festal <dfestal@redhat.com> * Also rename the attributes Signed-off-by: David Festal <dfestal@redhat.com> * Use a constant for the `workspace` prefix string Signed-off-by: David Festal <dfestal@redhat.com> * Fix formatting as requested by @sunix Signed-off-by: David Festal <dfestal@redhat.com> * Use a constant for the `isDevMachine` macro name Signed-off-by: David Festal <dfestal@redhat.com> * Add unit tests for `workspaceIdWithoutPrefixè and `isDevMachine` macros Signed-off-by: David Festal <dfestal@redhat.com> * Another requested formatting fix Signed-off-by: David Festal <dfestal@redhat.com> * Make new tests clearer Signed-off-by: David Festal <dfestal@redhat.com> * yet another formatting request Signed-off-by: David Festal <dfestal@redhat.com> * Respect the original order of imports Signed-off-by: David Festal <dfestal@redhat.com> * remove unnecessary `toString()` Signed-off-by: David Festal <dfestal@redhat.com> * use a lowercase `S` in the `server-` prefix Signed-off-by: David Festal <dfestal@redhat.com> * Multi-container workspace Support (eclipse-che#5110) * Fix 2 NPE that prevented using *non-dev* additional machines In the context of https://issues.jboss.org/browse/CHE-175 Signed-off-by: David Festal <dfestal@redhat.com> * Name openshift resources based on the machine name for non-dev machines This fixes https://issues.jboss.org/browse/CHE-259 and https://issues.jboss.org/browse/CHE-258 Signed-off-by: David Festal <dfestal@redhat.com> * Fix failing Traeffik tests... ... by: - adding the new `CHE_IS_DEV_MACHINE` env variable in tests - pulling up all the `CustomServerEvaluationStrategy` features in an abstract `BaseServerEvaluationStrategy` (which all other Che strategies extend) and have the `CustomServerEvaluationStrategy` class simply extend this `BaseServerEvaluationStrategy`. Signed-off-by: David Festal <dfestal@redhat.com> * Fix tests in the LocalDockerEvaluationStrategy... ... by correctly using the boolean attribute to manage the new use-case introduced by @fbenoit in master. Signed-off-by: David Festal <dfestal@redhat.com> * Replace OSIO-specific `single-port` strategy by `docker-local-custom` This fixes redhat-developer/rh-che#113 Signed-off-by: David Festal <dfestal@redhat.com>

l0rd approved these changes May 22, 2017

View reviewed changes

sunix reviewed May 22, 2017

View reviewed changes

sunix reviewed May 24, 2017

View reviewed changes

l0rd added the target/branch Indicates that a PR will be merged into a branch other than master. label May 24, 2017

davidfestal force-pushed the JBOSS-CHE-175 branch 2 times, most recently from 6ccb963 to d3cfa39 Compare May 29, 2017 12:19

davidfestal changed the base branch from openshift-connector to openshift-connector-rebased May 31, 2017 09:16

davidfestal force-pushed the JBOSS-CHE-175 branch from d3cfa39 to 92f4a43 Compare May 31, 2017 11:16

sunix force-pushed the openshift-connector-rebased branch 2 times, most recently from 607c92f to 8cb6d70 Compare June 2, 2017 20:50

davidfestal force-pushed the JBOSS-CHE-175 branch from 92f4a43 to fbd6941 Compare June 14, 2017 13:11

davidfestal changed the title ~~Multi-container workspace Support~~ [MERGE-AFTER-PR-5366] Multi-container workspace Support Jun 14, 2017

davidfestal force-pushed the JBOSS-CHE-175 branch 2 times, most recently from 4e437ee to 8b5cbbb Compare June 14, 2017 13:56

davidfestal mentioned this pull request Jun 15, 2017

Factorize code of ServerEvaluationStrategy classes, to use the Custom strategy as the basis of other strategies #5366

Merged

davidfestal force-pushed the JBOSS-CHE-175 branch from 8b5cbbb to 5a41c00 Compare June 15, 2017 09:52

davidfestal added 2 commits June 15, 2017 12:28

Fix 2 NPE that prevented using *non-dev* additional machines

115dce1

In the context of https://issues.jboss.org/browse/CHE-175 Signed-off-by: David Festal <dfestal@redhat.com>

Name openshift resources based on the machine name for non-dev machines

d8c5b2e

This fixes https://issues.jboss.org/browse/CHE-259 and https://issues.jboss.org/browse/CHE-258 Signed-off-by: David Festal <dfestal@redhat.com>

davidfestal force-pushed the JBOSS-CHE-175 branch from 5a41c00 to d8c5b2e Compare June 15, 2017 10:43

benoitf added the kind/enhancement A feature request - must adhere to the feature request template. label Jun 16, 2017

sunix merged commit 12f2db7 into eclipse-che:openshift-connector-rebased Jun 16, 2017

sunix mentioned this pull request Jun 29, 2017

Openshift connector improvements #5052

Merged

1 task

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[MERGE-AFTER-PR-5366] Multi-container workspace Support #5110

[MERGE-AFTER-PR-5366] Multi-container workspace Support #5110

davidfestal commented May 16, 2017 •

edited

Loading

codenvy-ci commented May 16, 2017

davidfestal commented May 16, 2017

l0rd left a comment

l0rd May 22, 2017

l0rd May 22, 2017

l0rd commented May 22, 2017

sunix May 22, 2017

davidfestal May 22, 2017

l0rd May 22, 2017

davidfestal commented May 23, 2017

sunix May 24, 2017

l0rd May 24, 2017

l0rd May 24, 2017

davidfestal commented May 24, 2017

davidfestal commented May 24, 2017

sunix commented May 24, 2017 •

edited

Loading

davidfestal commented May 24, 2017

davidfestal commented May 29, 2017

l0rd commented May 30, 2017

davidfestal commented May 31, 2017 •

edited

Loading

davidfestal commented Jun 15, 2017

benoitf commented Jun 16, 2017

davidfestal commented Jun 16, 2017

[MERGE-AFTER-PR-5366] Multi-container workspace Support #5110

[MERGE-AFTER-PR-5366] Multi-container workspace Support #5110

Conversation

davidfestal commented May 16, 2017 • edited Loading

What does this PR do?

What issues does this PR fix or reference?

codenvy-ci commented May 16, 2017

davidfestal commented May 16, 2017

l0rd left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

l0rd commented May 22, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

davidfestal commented May 23, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

davidfestal commented May 24, 2017

davidfestal commented May 24, 2017

sunix commented May 24, 2017 • edited Loading

davidfestal commented May 24, 2017

davidfestal commented May 29, 2017

l0rd commented May 30, 2017

davidfestal commented May 31, 2017 • edited Loading

davidfestal commented Jun 15, 2017

benoitf commented Jun 16, 2017

davidfestal commented Jun 16, 2017

davidfestal commented May 16, 2017 •

edited

Loading

sunix commented May 24, 2017 •

edited

Loading

davidfestal commented May 31, 2017 •

edited

Loading