-
Notifications
You must be signed in to change notification settings - Fork 71
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
e2e: use busybox to reduce e2e-test execution time #1605
e2e: use busybox to reduce e2e-test execution time #1605
Conversation
ca42c56
to
fc9b4b8
Compare
I knew @sudharshanibm3 and @stevenhorsman did many investigation on this unstable nginx pod, any comments on that? |
Hi @huoqifeng |
I still find it weird that Do we have any suspicion what a root cause might be, maybe the way a specific test is written? the linked issue above doesn't make it look like a well-known issue with library/nginx. I would at least suggest documenting the behaviour as a bug-issue "Can't use nginx in e2e tests" or sth and provided context, w/ a link to this and previous discussions. |
b0b5bf5
to
8549685
Compare
Yeah, it looks like not a well-known issue with library/nginx, I will try to do some debug on the failed test case, maybe we can pass the nginx issue with tiny changes. |
71ab108
to
a938e6a
Compare
We discussed this on the community call and the suggestion is that whilst we don't want to hide and lose the nginx issue, having a range of tests that occasionally fail isn't helpful either, so I propose that we swap the defaultimages for tests to busybox, but at the same time we add an nginx specific test, maybe using the approach Magnus mentioned: #1450 (comment) to add multiple replicas to get this reliable. |
c319900
to
b0c3f24
Compare
So this pr still valid that we replace replace nginx image with busybox image, I would like skip the new added deployment nginx test cases now, enable them after we have confidential-containers/guest-components#404 in kata side. WDYT @stevenhorsman @mkulke @wainersm @bpradipt ? |
So the kata fix should be merged into CAA very soon (#1632), so I think my suggested course, based on discussions last week
I'm not sure if that makes sense and is a reasonale plan? |
sounds good. maybe we can leave the simplepod test using nginx and switch to busybox for others, if just to reduce test execution time. A specific multi-layer test w/ a contrived image should be part of image-rs's unit tests, I think, we might not need it in CAA. |
a1d4aeb
to
9f2479b
Compare
9f2479b
to
a293a49
Compare
@mkulke @stevenhorsman I updated this PR please help to check again:
|
is |
We didn't have a test case to create one deployement with replicates, from this view, I prefer it is a new valid case? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hey DaLi, thanks a lot for this, it's looking pretty good. I have a few questions and comments:
0. I initially thought that refactoring some of the generic methods in nginx_deployment_test into a common file e.g. asssessment_runner_test.go would be better as they could be re-used, but on reflection I think you've done the correct thing and we have refactor out later if we want to re-used them rather than doing it straight away, so no need to change here
- Should we add the
doTestNginxDeployment
test to the other cloud-providers e.g. ibmcloud and azure, or where you thinking about doing that in a future PR instead? I think it's a good test, so don't want to see it forgotten - What is you reason for keeping createSimplePod test using nginx, given that we have the new nginx deployment test which more effectively tests the multi-layers issue? If we could switch it over to busybox (to match deleteSimplePodTest for example) then it should reduce the complexity of that test case as it only deals with a single layer image and would allow us to remove some of the extra code like
newNginxPod
andnewNginxPodWithName
which are only used in that single case? - I think there is an argument that the PR would be 'cleaner' if it was two separate commits - one for switching over nginx to busybox and another that adds the nginx deployment test, but this isn't a blocker
Sorry if I'm not coming across as being helpful in this review, or overly picking, I think the PR is good and none of these are blockers for merging, just thoughts that I had that I wanted to share and hear your comments on. Thanks!
- change to check container status for pod with running and testcommands. - use busybox to reduce test execution time - reduce WAIT_POD_RUNNING_TIMEOUT from 900 to 600 fixes for confidential-containers#1450 Signed-off-by: Da Li Liu <liudali@cn.ibm.com>
a293a49
to
71a2eb0
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks great, thanks for the updates!
Added this test cases to all provider.
Use busybox for
Split the codes to two commits.
Thanks @stevenhorsman good comments, I updated this PR. |
e8f86c8
to
3372f4a
Compare
- add a new test case create one nginx deployment with 2 replicas Signed-off-by: Da Li Liu <liudali@cn.ibm.com>
3372f4a
to
8886062
Compare
CI build passed when replicas of nginx deployment test is 2, it is failed when replicas is 5 it maybe caused by the limit resources. |
part of #1450
Signed-off-by: Da Li Liu liudali@cn.ibm.com