tracetest server install issue on kubernetes

[BartNetJS]

executing "tracetest server install"
select kubernetes

pod does not start with the log:

cannot read config file: open /app/config/config.yaml: no such file or directory
2023/05/10 15:54:13 Config file used: /app/config/config.yaml
2023/05/10 15:54:37 Config file used: /app/config/config.yaml

The reason is that config file is named tracetest-config1470662883 in the config map
It should be config.yaml

As a workaround I edited the CM with the correct config names.
The same for provisioning.

see the cm:

apiVersion: v1
kind: ConfigMap
metadata:
name: tracetest
namespace: tracetest
uid: 0b36545b-af6f-47ab-b483-8a0fcdfe16af
resourceVersion: '3484432'
creationTimestamp: '2023-05-10T15:53:31Z'
managedFields:
- manager: helm
operation: Update
apiVersion: v1
time: '2023-05-10T15:53:31Z'
fieldsType: FieldsV1
fieldsV1:
f:data: {}
- manager: kubectl-replace
operation: Update
apiVersion: v1
time: '2023-05-10T15:53:32Z'
fieldsType: FieldsV1
fieldsV1:
f:data:
f:tracetest-config1470662883: {}
f:tracetest-provisioning4237937856: {}
selfLink: /api/v1/namespaces/tracetest/configmaps/tracetest
data:
tracetest-config1470662883: |
postgres:
host: tracetest-postgresql
user: tracetest
password: not-secure-database-password
port: 5432
dbname: postgres
params: sslmode=disable
tracetest-provisioning4237937856: |+

---
type: Config
spec:
  analyticsEnabled: true
---
type: PollingProfile
spec:
  name: Custom Profile
  strategy: periodic
  default: true
  periodic:
    timeout: 2m
    retryDelay: 3s

binaryData: {}

[kdhamric]

Glad you had a workaround. Checking with the team to see if we can get it fixed.

[schoren]

hey @BartNetJS, I'm trying to reproduce this issue without luck.
I'm using tracetest v0.11.8 (just released).

I tried to do a full install: (with all dependencies)

also the more basic, only tracetest option:

I'm also a bit curios about the filename tracetest-config1470662883 in the config map you reported. That looks kind of like a temp filename, or one of those you get when doing kubectl edit. IIRC the intstaller does not use those methods, so I'm trying to figure out how it could end up looking like that.

Can you please confirm:

1. what version of tracetest cli are you running? (tracetest version)
2. what install option did you use? (full or basic)
3. what kind of k8s cluster are you using? (cloud, local, k3s, minikube, etc)
4. did you start with a clean state? Mainly, not having a pre existing tracetest namespace.

thanks!

[BartNetJS]

Hi @schoren

1. I'm now using the latest version: 0.11.8

3. kubernetes in docker
4. yesterday a clean install, today just upgrading

[jorgeepc]

Hey @BartNetJS @schoren we have two open issues regarding the same problem:

• Fix windows installer (Tracetest only) #2410
• Fix windows installer (Pokeshop demo) #2411

It occurs only on Windows.

[BartNetJS]

If I can help to tackle the issue, let me know

[kdhamric]

@BartNetJS Can you confirm you were installing on Windows when you had this issue?

[bunkrur]

I can say I am having the same issue from installing on Windows using v0.11.8.

tracetest server install completes, but the deployment on the pod is CrashLoopBackOff.

[BartNetJS]

@kdhamric I'm indeed installing on Windows 11

[kdhamric]

Very helpful - thanks @BartNetJS.

[kdhamric]

@bunkrur Were you able to use the same workaround @BartNetJS described earlier to get Tracetest install on Kubernetes on Windows? Did you get it successfully installed?

[bunkrur]

Yes, it worked. I also had to do the same workaround for a helm chart deployment.

[kdhamric]

@bunkrur Thanks.

@mathnogueira @schoren @danielbdias Thoughts on how difficult a correction / fix for this issue will be (or do we need time to dig into it to scope it?). There is a workaround, and we have a lot on our plate - just looking for a best guess so we can decide when to tackle this.