Skip to content

Instantly share code, notes, and snippets.

View AkashKumar7902's full-sized avatar
🎯
Focusing

Akash Kumar AkashKumar7902

🎯
Focusing
View GitHub Profile

3. Harness: PostgreSQL, Uyuni Workloads, and Observability

AkashKumar7902/uyuni#1

3.1 Purpose of This Phase

Section 2 established a storage-only baseline on local-path using repeated kubestr/fio runs. That baseline was necessary to isolate raw storage behavior before introducing database and application effects. The next step is to extend the measurement surface from block I/O into transaction-level and workload-level behavior while preserving the same evidence discipline: repeated runs, explicit metadata capture, raw logs, UTC timestamps, and structured summaries.

This phase therefore implements a reproducible benchmark harness that moves from storage-only measurement to storage-plus-database and storage-plus-application measurement. Concretely, it adds three capabilities on top of the Section 2 method: PostgreSQL benchmarking with pgbench, Uyuni workload benchmarking, and synchronized observability collection through Prometheus query windows and Grafana-compatible da

2. Benchmark Method, Execution, and Baseline Evidence

2.1 Scope and Goals

This section defines the storage-focused benchmark method used for evaluating Uyuni on RKE2 and records the first completed baseline run-set on the current storage backend. The purpose of this phase is to establish a reproducible control dataset before introducing PostgreSQL-level or Uyuni-level workload complexity. In other words, this phase measures the Kubernetes storage path itself, so that later performance changes can be attributed more confidently to the storage provider rather than to application behavior.

The goals of this section are threefold:

  1. Define a repeatable, storage-only experiment that can be rerun under identical conditions.
  2. Document the execution logic, automation steps, and generated artifacts in a way that supports reproducibility.
This file has been truncated, but you can view the full file.
apiVersion: v1
items:
- apiVersion: v1
kind: Pod
metadata:
annotations:
cni.projectcalico.org/containerID: 00aff01014384b840c8cf7a95c0acd61b2d9974088f7a515f4e2a237258acbbe
cni.projectcalico.org/podIP: 10.42.0.56/32
cni.projectcalico.org/podIPs: 10.42.0.56/32
prometheus.io/path: /metrics
@AkashKumar7902
AkashKumar7902 / harbor-proposal.md
Last active February 2, 2024 15:41
LFX'24 PROJECT PROPOSAL (HARBOR)
@AkashKumar7902
AkashKumar7902 / proposal.md
Last active February 2, 2024 08:54
LFX'24 PROJECT PROPOSAL (Harbor)