Release v7

[mraible]

Overview of the feature request

This ticket is to track the release of JHipster 7 final. Things I'm looking into:

• Couchbase. The last time I checked, Spring MVC + Couchbase was working, but WebFlux was not. If that’s still the case, I think it’s a good idea to disable Couchbase as an option for reactive. Upgrade Couchbase support for Spring Boot 2.4 #14184 and Disable Couchbase and move WebFlux and Infinispan out of beta #14186
• yeoman-generator@5 Update to yeoman-generator@5 #14064
• Test and make sure the blog post I wrote on Reactive Java Microservices still works. jhipster jdl reactive-ms fails e2e tests #14196
• Deploy the result of the previous blog post to Google Cloud and make sure our k8s generator works.
• Remove traefik gateway option (Remove traefik support #14233)

Motivation for or Use Case

Quality

[mraible]

I'll update this comment with the status as I address the issues above.

• ❌ ngx-couchbase compile error
• ❌ webflux-couchbase same as above.
• ❌ jhipster jdl reactive-ms fails with Protractor jhipster jdl reactive-ms fails e2e tests #14196

Update: Couchbased removed in #14186 so the first two are no longer an issue. The 3rd item is still an issue and also might have problems on Windows.

[mraible]

It'd be cool if we could clean this up at the end of jhipster jdl reactive-ms:

You can launch all your infrastructure by running : docker-compose up -d
Congratulations, JHipster execution is complete!
Sponsored with ❤️  by @oktadev.
INFO! Generator docker-compose child process exited with code 0
Congratulations, JHipster execution is complete!
Congratulations, JHipster execution is complete!

I think the following would be better:

You can launch all your infrastructure by running : docker-compose up -d
Congratulations, JHipster execution is complete!
Sponsored with ❤️  by @oktadev.

Definitely a stretch goal.

[mraible]

Added a bug bounty so y'all can expense beers on Okta after the release. 😄

[mshima]

I'll update this comment with the status as I address the issues above.

• ❌ ngx-couchbase EJS error with primary keys
• ❌ webflux-couchbase same as above.
• ❌ jhipster jdl reactive-ms fails with Protractor:

I tried changing the gateway app to use Cypress and the errors still happen. I'm not sure why. I'll dig in tomorrow (Friday).

@mshima What are we doing wrong for primaryKey?

Not sure. I take a look.

[mshima]

@mraible fixed couchbase primaryKey reference in your mongodb branch. Realized later that mongodb branch was not couchbase 😆 .

I can setup an environment to test if there are more problems later.

edit:
Fixed the generation problem, but there is another problem

'dependencies.dependency.version' for com.couchbase.client:encryption:jar is missing. @ line 159, column 21

https://github.com/hipster-labs/jhipster-daily-builds/runs/2039400380?check_suite_focus=true

[mraible]

I just realized that Couchbase + Spring Boot 2.4 is still pending in this PR.

[pascalgrimaud]

@mshima : should we add yeoman 5 in the list ?

[mshima]

@pascalgrimaud added. Hopefully I can finish the PR today.

[pascalgrimaud]

Just crazy!

Ok so I won't do the final v7 release this weekend, so probably the next weekend, to give time to finish the last items here.
I can only do release saturday or sunday as I'm too busy with my day work at the moment.

[gzsombor]

Is it possible to create a release just from the JHipster-bom/framework? A 7.0.0-beta-2 or something like that? Currently, the generator use 7.0.0-SNAPSHOT dependency, which makes local testing a bit fragile.

[mraible]

I believe @jdubois might've fixed things so the SNAPSHOT build is published nightly.

[jdubois]

It should work automatically but I didn't test -> please mention/contact me if you don't see the snapshots

[pascalgrimaud]

Is it possible to create a release just from the JHipster-bom/framework? A 7.0.0-beta-2 or something like that? Currently, the generator use 7.0.0-SNAPSHOT dependency, which makes local testing a bit fragile.

@gzsombor : you need to do these steps

• go into your local jhipster-bom (which should be a fork)
• update your fork
• mvn clean install -Dgpg.skip

It can be easily done with an alias. It took less than 15sec, if you decide to skip tests:

[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time:  13.389 s
[INFO] Finished at: 2021-03-06T09:46:34+01:00
[INFO] ------------------------------------------------------------------------

[pascalgrimaud]

similar to improvement suggested by Matt, after generating full microservices applications, there are some changes, which are not correctly commit with git

[mshima]

@pascalgrimaud can you confirm if the changes are only json files?

[pascalgrimaud]

Yes, package.json, package lock and yo rc, if I remember well

[mshima]

Yes, package.json, package lock and yo rc, if I remember well

I was digging into it to make sure it wasn't related to the generator5 update.
And saw that package.json changes is pretty big.
After a couple of hours on this, just realized the problem is prettier-plugin-packagejson version mismatch.
prettier-plugin-packagejson stopped sorting scripts due to a change at sort-package-json keithamus/sort-package-json#206.

So the generator generates using prettier-plugin-packagejson from generator-jhipster node_modules.
After generated and install, husky task executes prettier using prettier-plugin-packagejson from destination node_modules.
prettier-plugin-packagejson is not added the destination package.json, so the version from generator-jhipster@7.0.0.beta.1.

Including prettier-plugin-packagejson in the destination package.json may fix this issue for the package.json, yo-rc.json is a different.

[pascalgrimaud]

I'm not sure for yo-rc as I'm not with my computer right now :-p
Nice catch anyway for package json

[mshima]

yo-rc.json is due to jdl writing it disc and generator doesn't change it, since it hasn't been changed, prettifier is not executed.
I will fix this.

[egvimo]

Would you mind to release 7.0.0.beta.2 meanwhile? The milestone is on 100% and there are some fixes, which I really would like to migrate from the beta.1.

[mshima]

@pascalgrimaud even with the PR there will be a diff in .yo-rc.json. Docker-compose writes to it after it was committed to git.

[pascalgrimaud]

@mshima : no worry about yo rc, it's minor

[pascalgrimaud]

@egvimo : the release beta.2 should be for the next week-end but I can't promise, as it's open source and we work for JHipster during our freetime. I'll do my best, just be patient :-)

[DanielFran]

Added option to remove traefik support

[mraible]

Couchbase is removed so I'm only concerned with jhipster jdl reactive-ms and deploying it with k8s now.

[mraible]

I'm trying to deploy the reactive-ms stack on GCP with K8s now. When I run ./kubectl-apply.sh -f, I'm getting the following error:

Unable to connect to the server: dial tcp 35.233.152.164:443: i/o timeout

I was able to deploy my Docker images directly to my registry with commands like:

./gradlew bootJar -Pprod jib -Djib.to.image=gcr.io/ionic-social-286904/gateway

[mraible]

I was finally able to make things work after creating a new project and cluster. We might want to update our GCP docs with more detailed steps.

After running ./kubectl-apply.sh -f, things were looking good.

Applying default `kubectl apply -f`
configmap/application-config created
secret/registry-secret created
service/jhipster-registry created
statefulset.apps/jhipster-registry created
deployment.apps/blog created
persistentvolumeclaim/blog-neo4j-pvc created
secret/blog-neo4j created
statefulset.apps/blog-neo4j created
service/blog-neo4j created
service/blog created
deployment.apps/gateway created
persistentvolumeclaim/gateway-postgresql-pvc created
secret/gateway-postgresql created
deployment.apps/gateway-postgresql created
service/gateway-postgresql created
service/gateway created
deployment.apps/store created
configmap/store-mongodb-config created
configmap/store-mongodb-init created
statefulset.apps/store-mongodb created
service/store-mongodb created
service/store created

10 minutes later and it seems some containers failed to start.

$ kubectl get pods
NAME                                 READY   STATUS     RESTARTS   AGE
blog-69c885659-mjfs6                 0/1     Init:0/1   0          11m
blog-neo4j-0                         0/1     Pending    0          11m
gateway-d76f7bff6-9pc2g              0/1     Pending    0          11m
gateway-postgresql-f9bbb4b65-lp75b   0/1     Pending    0          11m
jhipster-registry-0                  1/1     Running    0          11m
jhipster-registry-1                  1/1     Running    0          11m
store-79fd4f88c4-9xbw8               0/1     Pending    0          11m
store-mongodb-0                      0/1     Pending    0          11m

Screenshot of workloads in Kubernetes Engine on GCP:

Any ideas @saturnism?

[atomfrede]

Can you check how the resource limits and requests are configured for blog-neo4j? Looks like the nodes are to small for the requested amount of memory or cpu.

[mraible]

Yes, I think that's it. I need to use better machines. Maybe I can specify the machine type when creating the cluster?

[pascalgrimaud]

I was thinking we are closed to v7 but it seems a lot of things are broken - https://github.com/hipster-labs/jhipster-daily-builds
Need to wait before release v7...

[DanielFran]

@pascalgrimaud I am confident the update to spring-boot 2.4.4 will fix it. It seems the issue started after I pushed the override of neo4j to 6.0.6...

[DanielFran]

@pascalgrimaud In fact, looking now at the chronology of the merges, it might be that the issue started when it was merged #14365 or #14364

I already take a look and did not identify what could be the issue...

[mraible]

I tried creating a new cluster on GCP with bigger machines.

$ gcloud container clusters create jhipster \
--zone us-central1-a \
--machine-type n1-standard-4 \
--enable-autorepair \
--enable-autoupgrade

Then, I ran ./kubectl-apply.sh -f and deployed. For some reason, it seems there's an issue pulling my images.

$ kubectl get pods
NAME                                 READY   STATUS             RESTARTS   AGE
blog-69c885659-44phh                 0/1     ImagePullBackOff   0          14m
blog-neo4j-0                         1/1     Running            0          14m
gateway-d76f7bff6-jgwxb              0/1     ImagePullBackOff   0          14m
gateway-postgresql-f9bbb4b65-9r4k5   1/1     Running            0          14m
jhipster-registry-0                  1/1     Running            0          14m
jhipster-registry-1                  1/1     Running            0          14m
store-79fd4f88c4-cngjn               0/1     ImagePullBackOff   0          14m
store-mongodb-0                      1/1     Running            0          14m
store-mongodb-1                      1/1     Running            0          13m
store-mongodb-2                      1/1     Running            0          13m

I tried kubectl delete pod blog-69c885659-44phh (and others), then running ./kubectl-apply.sh -f again, but no luck. 🤷‍♂️

UPDATE:

Figured it out. When I ran jhipster k8s, I was pointing to my old project, which I've since deleted. kubectl describe pod blog-69c885659-wnwrt gave me the answer.

  Normal   Pulling    2m5s (x4 over 3m34s)  kubelet            Pulling image "gcr.io/ionic-social-286904/blog"
  Warning  Failed     2m5s (x4 over 3m34s)  kubelet            Failed to pull image "gcr.io/ionic-social-286904/blog": rpc error: code = Unknown desc = Error response from daemon: Get https://gcr.io/v2/ionic-social-286904/blog/manifests/latest: denied: Project ionic-social-286904 has been deleted.
  Warning  Failed     2m5s (x4 over 3m34s)  kubelet            Error: ErrImagePull
  Normal   BackOff    113s (x5 over 3m7s)   kubelet            Back-off pulling image "gcr.io/ionic-social-286904/blog"

I tried deleting pods and re-deploying, but that didn't work. Deleting and re-creating my cluster did.

[mshima]

The problem is with mysql, not sure if the driver or the docker image.
Every test that I've looked at has mysql server.

[DanielFran]

Image has been updated 10 days ago, so it cannot be that.
I was thinking about the driver but did not found what commit would be the origin of the issue...

[mraible]

Does anyone know how to enable HTTPS on my apps that are deployed to GCP? I have everything running, but it's all over HTTP, which won't work with Okta.

[mshima]

@DanielFran I've triggered an old commit that succeed and if failed with the same error.
It should be related to jhipster-bom or an outside problem.

[mraible]

I figured out how to do HTTPS for my K8s cluster with @saturnism's external load balancing guide. Now that I've deployed everything successfully, I'm 👍 for releasing JHipster 7!

Of course, I created my apps last night, so recent changes might've broken things. :)

[pascalgrimaud]

Ok, I'll ask a code freeze in the mailing list so let few days for daily builds

[mraible]

I was able to get HTTP -> HTTPS working on my K8s cluster using the following code.

// Force HTTPS
http.redirectToHttps(redirect -> redirect
    .httpsRedirectWhen(e -> e.getRequest().getHeaders().containsKey("X-Forwarded-Proto"))
);

[pascalgrimaud]

Closed by:

• https://www.jhipster.tech/2021/03/21/jhipster-release-7.0.0.html
• https://github.com/jhipster/generator-jhipster/releases/tag/v7.0.0

[atomfrede]

Thanks @pascalgrimaud and everyone of the team. It's a pleasure to work with you! When will we have our virtual drink? 🍻 😁

[pascalgrimaud]

We should organize it, and discuss about v8 :-)