Use minishift in CI #214

machacekondra · 2019-03-22T08:33:35Z

What this PR does / why we need it:

Which issue(s) this PR fixes (optional, in fixes #<issue number>(, fixes #<issue_number>, ...) format, will close the issue(s) when PR gets merged):
Fixes #

Special notes for your reviewer:

Release note:

vatsalparekh

This PR is needed to run changes of #212 related to openshift templates

tests/playbooks/e2e.yaml

]

mmazur · 2019-04-25T13:38:23Z

I've bumped the kubevirt_vm call that times out to have a timeout value of 5 minutes instead of 2 minutes and it still fails. Looks like something else is wrong. Let's see…

mmazur · 2019-04-26T13:38:27Z

I'm pretty sure I managed to get the vms running by messing a bit with how kubevirt is getting deployed, but now I've hit an interesting problem. When an ephemeral vm exists and is running (I'm guessing this isn't just limited to ephemeral vms), the task below should return success with changed==false.

kubevirt_vm:
  state: present
  name: xyz
  ephemeral: yes

And it does on kubevirt 1.6.0 running on vanilla k8s 1.13.3. But it doesn't on same kubevirt running on latest minishift (or not so latest minishift). And this is the exact issue currently reported by travis. I'm suspecting some weird behavior on the openshift lib layer.

mmazur · 2019-04-26T14:39:26Z

Blocked by: ansible/ansible#55817

mmazur · 2019-04-29T09:33:12Z

Ugh, so openshift 3.11 ships kubernetes 1.11 and there's a bug in that k8s that only got fixed in 1.12 (kubernetes/kubernetes#67562), which is what we're hitting here.

I'll see if I can get the ansible/k8s people to add a workaround for this (ideally before 2.8 ships).

mmazur · 2019-04-29T14:36:13Z

😠 !#@$!^#$ 😡

Works… kind of. I'm pretty sure this will be even more flaky than the minkube environment, so expect random stuff to time out for no reason. Might be a problem for us, might not. If it is, we can investigate something other than travis at some later time.

For now, @machacekondra @pkliczewski @vatsalparekh please review and merge.

pkliczewski · 2019-04-29T14:38:39Z

.gitmodules

@@ -1,3 +1,4 @@
 [submodule "ci"]
 	path = ci
-	url = https://github.com/fabiand/traviskube
+	url = https://github.com/mmazur/traviskube


I understand there are bugs. Please paste reference or describe why we are using our fork.

My changes are very trivial, and update here and there. The only reason is the RH Summit. Fabian won't be looking at low prio PRs for some time and I didn't want to get stuck on this. As soon Fabian is back to his usual responsive self I'm planning on getting my changes merged and reverting this back to his repo.

Maybe we can move it under kubevirt org

pkliczewski · 2019-04-29T14:39:32Z

.travis.yml

+    - name: "minishift"
+      env:
+        - CPLATFORM=minishift
+        - KUBEVIRT=v0.16.0


v0.16.2 - latest released

0.16.1 and 0.16.2 are tagged as pre–release, which is why I didn't use them. Dunno if those tags can be ignored or if they actually mean something.

There is v0.17.0 available please change.

pkliczewski · 2019-04-29T14:40:52Z

tests/playbooks/kubevirt_vm.yml

      with_items:
        - { name: 'eph1', state: 'present' }
        - { name: 'eph1', state: 'running' }
        - { name: 'eph2', state: 'present' }
        - { name: 'eph2', state: 'running' }
      register: result
-      failed_when: result.changed != false or result.kubevirt_vm.status.phase != 'Running'
+      failed_when: result.kubevirt_vm.status.phase != 'Running'
+#     failed_when: result.changed != false or result.kubevirt_vm.status.phase != 'Running'


why do we need this commented line?

There's a 4 line explanation of what this is for at the start of the playbook.

I did it this way on purpose. If I'm making a temporary workaround which I want to get removed as soon as possible, then I want to very clearly mark all the places in the code where the workaround applies. The only chance for someone to update it in a few months so that full tests are utilized is to make it an eye sore and hard to not notice should anyone look at this file.

You know how this stuff works. If I hide these lines in git, then no one will ever bother to try to fish out the originals and figure out what to do with them, so the "workaround" will just become the permanent tests until the heat death of the universe. But if the originals are still there in plain sight, there's a chance…

pkliczewski · 2019-04-29T14:41:03Z

tests/playbooks/kubevirt_vm.yml

      with_items:
        - { name: 'eph1', state: 'stopped' }
        - { name: 'eph1', state: 'absent' }
        - { name: 'eph2', state: 'stopped' }
        - { name: 'eph2', state: 'absent' }
      register: result
-      failed_when: result.changed != false or 'spec' in result.kubevirt_vm or 'status' in result.kubevirt_vm
+      failed_when: ('spec' in result.kubevirt_vm or 'status' in result.kubevirt_vm)
+#     failed_when: result.changed != false or 'spec' in result.kubevirt_vm or 'status' in result.kubevirt_vm


pkliczewski · 2019-04-29T14:41:32Z

tests/playbooks/kubevirt_vm.yml

      with_items:
        - { name: 'vm1', state: 'stopped' }
        - { name: 'vm1', state: 'present' }
        - { name: 'vm2', state: 'stopped' }
        - { name: 'vm2', state: 'present' }
      register: result
-      failed_when: result.changed != false or 'status' in result.kubevirt_vm
+      failed_when: ('status' in result.kubevirt_vm)
+#     failed_when: result.changed != false or 'status' in result.kubevirt_vm


I see there are more. We use version control not to keep things like this.

machacekondra · 2019-04-29T16:37:13Z

I would rather prefer stable tests without testing templates, then unstable env with templates testing.

mmazur · 2019-04-30T08:13:27Z

It's my guess that the reliability issues we've had with the test env were due to k8s/kubevirt and non-accelarated vms in particular being rather heavy, so whenever there was a cpu shortage we got random timeouts.

And since k8s is lighter than openshift I suspect that might be more of a problem.

But I don't know for sure and this patch is already done. If it turns out to be a problem, we can revert this easily or consider moving away from travis. Or maybe it'll be enough to increase a few timeout values. I'll keep an eye on it going forward.

machacekondra · 2019-05-07T08:21:22Z

test.sh

-echo "$(minikube ip) minikube" | sudo tee -a /etc/hosts
-
-## No HW virt
-kubectl create configmap -n kubevirt kubevirt-config --from-literal debug.useEmulation=true


We don't need this anymore?

Ah, ok, you moved most of this to your repo.

mmazur · 2019-05-21T09:39:36Z

I'll open a separate PR that adds jenkins support.

kubevirt-bot added the size/S label Mar 22, 2019

machacekondra force-pushed the use_minishift branch from 0b61be6 to fc5b5f2 Compare March 22, 2019 12:48

kubevirt-bot added size/M and removed size/S labels Mar 22, 2019

machacekondra force-pushed the use_minishift branch 2 times, most recently from 04cd320 to 3647352 Compare March 22, 2019 12:56

kubevirt-bot added size/S and removed size/M labels Mar 22, 2019

machacekondra force-pushed the use_minishift branch 7 times, most recently from 675f329 to 22f253c Compare March 22, 2019 19:37

kubevirt-bot added size/M and removed size/S labels Mar 22, 2019

machacekondra force-pushed the use_minishift branch from 22f253c to 2d0e8e8 Compare March 22, 2019 19:51

vatsalparekh suggested changes Apr 22, 2019

View reviewed changes

tests/playbooks/e2e.yaml Outdated Show resolved Hide resolved

] Outdated Show resolved Hide resolved

vatsalparekh mentioned this pull request Apr 22, 2019

Changes in playbooks for automation tests #212

Closed

machacekondra force-pushed the use_minishift branch 2 times, most recently from 6d87a67 to 7fbb234 Compare April 23, 2019 07:12

kubevirt-bot added size/S and removed size/M labels Apr 23, 2019

machacekondra force-pushed the use_minishift branch from 7fbb234 to d5026e8 Compare April 23, 2019 07:12

mmazur force-pushed the use_minishift branch from 3352188 to ce91fb6 Compare April 26, 2019 11:14

kubevirt-bot added size/M and removed size/S labels Apr 26, 2019

mmazur force-pushed the use_minishift branch from ce91fb6 to fc727d1 Compare April 26, 2019 11:18

Switch away from Fabian's traviskube repo

ed6b332

mmazur force-pushed the use_minishift branch from fc727d1 to b08f8fd Compare April 29, 2019 07:57

mmazur force-pushed the use_minishift branch 6 times, most recently from 1e83603 to 6f6c35b Compare April 29, 2019 14:07

pkliczewski reviewed Apr 29, 2019

View reviewed changes

vatsalparekh approved these changes May 7, 2019

View reviewed changes

pkliczewski approved these changes May 7, 2019

View reviewed changes

mmazur mentioned this pull request May 7, 2019

Add ENABLE_DATAVOLUMES to deploy-kubevirt fabiand/traviskube#40

Closed

pkliczewski closed this May 7, 2019

pkliczewski reopened this May 7, 2019

machacekondra commented May 7, 2019

View reviewed changes

mmazur force-pushed the use_minishift branch from 4f20112 to 435ca29 Compare May 7, 2019 09:53

mmazur added 3 commits May 7, 2019 11:56

Use minishift with latest kubevirt; also gather debug info

13b0575

Add workaround for os 3.x test env

3aa8c10

Correct pvc sizes to fit fetched images

1d9eb92

mmazur force-pushed the use_minishift branch from 435ca29 to 1d9eb92 Compare May 7, 2019 09:57

More updates

24222a4

mmazur force-pushed the use_minishift branch from 73f68ee to 24222a4 Compare May 7, 2019 12:43

mmazur closed this May 21, 2019

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Use minishift in CI #214

Use minishift in CI #214

machacekondra commented Mar 22, 2019

vatsalparekh left a comment

mmazur commented Apr 25, 2019

mmazur commented Apr 26, 2019

mmazur commented Apr 26, 2019

mmazur commented Apr 29, 2019

mmazur commented Apr 29, 2019

pkliczewski Apr 29, 2019

mmazur Apr 30, 2019

vatsalparekh May 2, 2019

pkliczewski Apr 29, 2019

mmazur Apr 30, 2019

pkliczewski May 7, 2019

pkliczewski Apr 29, 2019

mmazur Apr 30, 2019

pkliczewski Apr 29, 2019

pkliczewski Apr 29, 2019

machacekondra commented Apr 29, 2019

mmazur commented Apr 30, 2019

machacekondra May 7, 2019

machacekondra May 7, 2019

mmazur commented May 21, 2019

Use minishift in CI #214

Use minishift in CI #214

Conversation

machacekondra commented Mar 22, 2019

vatsalparekh left a comment

Choose a reason for hiding this comment

mmazur commented Apr 25, 2019

mmazur commented Apr 26, 2019

mmazur commented Apr 26, 2019

mmazur commented Apr 29, 2019

mmazur commented Apr 29, 2019

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

machacekondra commented Apr 29, 2019

mmazur commented Apr 30, 2019

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mmazur commented May 21, 2019