pulsar manager didn't start postgres. So it didn't start to work with

[ton3r]

Describe the bug

If you have the problem that one will have a 404 Error in the WebConsole during Login on pulsar manager. I have installed it from

https://pulsar.apache.org/docs/3.1.x/helm-deploy/

During my debug, I found out that this is ok. Because the Backend didn't start.

After connecting to the shell of ird-pulsar-pulsar-manager I've seen in the tail -f pulsar_manager.log of the spring instance that it can't connect to postgres on 127.0.0.1. There wasn't any available.

Installing apt-get update && apt-get -y install procps lsof vim and I was possible to figure out that there was nothing listening on port 7750 or 5432

So I've created my own postgres instance somewhere, create the schema with this link and startet the helm install like

helm install \
    --values pulsar-ird/values.yaml \
    --set initialize=true \
    --set proxy.ports.http=8080 \
    --set proxy.securityContext.runAsUser=0 \
    --set proxy.securityContext.runAsGroup=0 \
    --set proxy.service.loadBalancerIP=192.168.77.101 \
    --set pulsar_manager.admin.user=postgres \
    --set pulsar_manager.admin.password=POSTGRES_PASSWORD \
    --set pulsar_manager.configData.URL=jdbc:postgresql://192.168.77.10:5432/pulsar_manager \
    --set pulsar_manager.service.loadBalancerIP=192.168.77.100 \
    --namespace pulsar \
    ird-pulsar apache/pulsar

I've to switch the proxy to 8080 because bind to 80 wasn't possible. Also the possible "solutions" setting the runAs* parameter to root didn't help. So I take port 8080.

After the helm command is finished, I connect to the shell again and tail -f pulsar_manager.log again. Look out for Hikari Logs. That went well.

After the spring service started up and was listening on 7750 I was possible to create an admin user!
Yes! One need this. It's not directly documented by the how tos!! Just during my investigations at least here in the issue board.

CSRF_TOKEN=$(kubectl get secret --namespace pulsar ird-pulsar-token-admin -o jsonpath="{.data.TOKEN}" | base64 -d)
curl \
    -H "X-XSRF-TOKEN: $CSRF_TOKEN" \
    -H "Cookie: XSRF-TOKEN=$CSRF_TOKEN;" \
    -H 'Content-Type: application/json' \
    -X PUT http://THE_MANAGER_IP:9527/pulsar-manager/users/superuser \
    -d '{"name": "admin", "password": "BirnenBurner123_", "description": "test", "email": "[email protected]"}'

Then you should be able to login with admin:BirnenBurner123_

And fun-fact : pulsar_manager.admin.user and pulsar_manager.admin.password are the db access credits (may be also some credits for other things..)

Bug Report 1 : In the end I am still struggling to set my desired internal IPs. They got ignored and always got new ip's on re-deploys.

Bug Report 2 : What is the definition of "production ready" ? And why uses the pulsar_manager by default on internal postgres instance (which is not starting up) in a non-stateful-set?

Bug Report 3: The helm chart is not compatible kubernetes api v1.27. It's using deprecated (beta) apis and webhooks from within prometheus

To Reproduce
As written up.

Expected behavior

• That a database is spawned up,
• that it's statefull and
• that I can set ip-addresses
• would be easier to find the spring logs from pulsar manager

Additional context
Deployed in a GKE 1.27 Cluster.

[mindesclabs]

I'm having the same issue with DB start...local postgresql DB inside manager pod is not running.

[ton3r]

Try to configure as describe an external Database. I've switched to installations on VM's. There is also the default an internal postgres (which really starts..) but on restart of the manager you loose your configured UI. But you can also configure an external database.

[mindesclabs]

Yeah..I'll have to go with external DB. I've managed to get running internal postgres DB by using newer manager image (helm is using 0.3 which does not load DB). But I can see your point. If no Volume is provisioned for the manager pod it is useless for production so only choice is to use external DB

[lhotari]

Fixed by #457