Week 11

(August 12, 2025 – August 18, 2025)

Scheduler Agent Work

This week, most of time was again focused on the Scheduler pod, as it is the core component that controls the execution of other agents. Without the scheduler running properly, dependent agents like Ojo and Nomos, Web etc… cannot start.

I continued testing, reviewing logs and updating configurations (such as changes in docker-entrypoint.k8s.sh and php-conf-fix.sh) to push the scheduler closer to a stable state.

Currently, the scheduler is more than partially working and it’s a step forward compared to previous weeks. And I was able to confirm some progress on the Web pod as well. Getting the UI locally is still in progress, but the debugging effort is paying off.

Here is the progress of both Scheduler and web Agents:

Scheduler Pod Logs Partially running
Web Pod Progress

This remains the main blocker in our microservices deployment, and once it is stable, the dependent agents should also run smoothly.

Kustomization Setup

Alongside the scheduler debugging, I spent time learning and working on Kustomization as per suggestion of Avinal. The goal here is to properly manage multiple Kubernetes YAMLs in a scalable way.

I’ve been going through examples of base and overlay (dev/prod) structures and planning how to fit them into our current k8s/ directory. This work is ongoing and I continue to refine the setup in the coming week. Here’s the structure I’ve built:

k8s/
├── base/
│   ├── 00-configmap.yaml
│   ├── 01-db-secret.yaml
│   ├── 02-db-service.yaml
│   ├── 03-database-persistentvolumeclaim.yaml
│   ├── 04-db-statefulset.yaml
│   ├── 05-etcd-service.yaml
│   ├── 06-etcd-persistentvolumeclaim.yaml
│   ├── 07-etcd-deployment.yaml
│   ├── 08-scheduler-service.yaml
│   ├── 09-service-account-scheduler.yaml
│   ├── 10-repository-persistentvolumeclaim.yaml
│   ├── 11-scheduler-deployment.yaml
│   ├── 12-agent-service.yaml
│   ├── 13-adj2nest-deployment.yaml
│   ├── 14-ununpack-deployment.yaml
│   ├── 15-wgetagent-deployment.yaml
│   ├── 16-copyright-deployment.yaml
│   ├── 17-nomos-deployment.yaml
│   ├── 18-ojo-deployment.yaml
│   ├── 19-web-service.yaml
│   ├── 20-web-deployment.yaml
│   ├── 21-monk-deployment.yaml
│   ├── 22-scanoss-deployment.yaml
│   ├── 23-delagent-deployment.yaml
│   ├── 24-cyclonedx-deployment.yaml
│   ├── 25-readmeoss-deployment.yaml
│   ├── configmap.yaml.example
│   ├── db-secret.yaml.example
│   ├── fo_delete.sh
│   └── kustomization.yaml
├── overlays/
│   ├── dev/
│   │   ├── patch-scheduler-deployment.yaml
│   │   ├── patch-db-secret.yaml
│   │   └── kustomization.yaml
│   ├── prod/
│   │   ├── patch-scheduler-deployment.yaml
│   │   ├── patch-db-secret.yaml
│   │   └── kustomization.yaml

Dockerfiles Simplification

Another point I discussed with mentors was the need to redo the Dockerfiles. The current Dockerfiles work, but they are heavy and harder to maintain. A simpler configuration will not only make builds faster but also reduce complexity for future development. This task is still in progress and I will keep iterating on it.

Collaboration and Mentorship

I have been actively collaborating with Avinal on scheduler debugging. We discussed different approaches, including curl-based communication and the possibility of shared resources for agents, but for now, we are focusing on getting the current setup running first before moving to optimal solutions.
I also reached out to Omar (who worked on FOSSology microservices in GSoC 2021) for clarifications. I’m still awaiting his response, but his insights will be important to unblock some parts of this work.

Meeting 1

(August 13, 2025)

Attendees:

In the mentor call this week, I shared updates on:

Partial progress with the Scheduler pod
Debugging work done on docker-entrypoint.k8s.sh and php-conf-fix.sh
Early testing of the Web pod with scheduler integration
Plan to simplify Dockerfiles

The discussion confirmed that for now we should continue with the curl-based scheduler approach, and later explore the optimal solution once things are stable.

Meeting 2

(August 14, 2025)

In the weekly community meeting call, I summarized my progress:

Spent most of the week debugging the Scheduler pod, and got it partially running
Worked on the Web pod, which now shows partial connectivity with the scheduler
Continued learning Kustomization for Kubernetes deployments
Mentioned the plan to redo Dockerfiles for cleaner configuration