Pavel Usanli
Projects

Kubernetes Utility Containers - DevOps Automation Images

Lightweight production-ready Docker utility containers for Kubernetes environments. Includes a PostgreSQL backup/restore image with AWS S3 integration and retention policies, and a kubectl + AWS CLI toolbox image designed for automation jobs such as ECR secret refresh, cluster maintenance, and GitOps workflows.

  • Docker
  • Kubernetes
  • PostgreSQL
  • AWS CLI
  • S3
  • kubectl
  • Alpine Linux
  • Shell Scripting
  • GitHub Actions
  • GHCR

A small set of utility container images meant to be dropped into Kubernetes CronJobs and Jobs for common ops tasks:

  • postgres-awscli - PostgreSQL backups/restores to S3 (plus retention)
  • kubectl-awscli - kubectlCopied + AWS CLI toolbox image for cluster automation

Both images are published to GitHub Container Registry (GHCR).
The easiest way to see all available tags/versions is to open the package page:

On those pages you’ll typically find:

  • latest (moving tag)
  • version tags (e.g. 9d4d16a)

For full usage docs and configuration options, check:

  • postgres-awscli README: README
  • kubectl-awscli README: README

What each image is for

1. postgres-awscli

Purpose: run database backups on a schedule (CronJob), and restores as an on-demand Job.

Typical capabilities (see README for the authoritative list):

  • pg_dump + psql (PostgreSQL client tools)
  • aws-cli
  • Backup mode (default): dump → compress → upload to S3 → retention
  • Restore mode: download from S3 → restore into PostgreSQL (optionally recreate DB)

Docs: see the latest postgres-awscli README for

  • required env vars (DB connection, S3 bucket/prefix, AWS region/creds/IAM role)
  • restore parameters (which key to restore, optional recreate behavior)
  • Kubernetes CronJob and Job manifests

Quick example (local backup run)

Variables and exact names can change - double check the README before copy/paste.

bash
docker run --rm \
  -e PGHOST=localhost \
  -e PGPORT=5432 \
  -e PGDATABASE=mydb \
  -e PGUSER=myuser \
  -e PGPASSWORD=mypassword \
  -e AWS_ACCESS_KEY_ID=xxx \
  -e AWS_SECRET_ACCESS_KEY=yyy \
  -e AWS_DEFAULT_REGION=eu-central-1 \
  -e S3_BUCKET=my-backups \
  -e S3_PREFIX=pg/mydb \
  ghcr.io/kalpak44/postgres-awscli:latest

Quick example (local restore run)

bash
docker run --rm \
  -e MODE=restore \
  -e PGHOST=localhost \
  -e PGPORT=5432 \
  -e PGDATABASE=mydb \
  -e PGUSER=myuser \
  -e PGPASSWORD=mypassword \
  -e AWS_ACCESS_KEY_ID=xxx \
  -e AWS_SECRET_ACCESS_KEY=yyy \
  -e AWS_DEFAULT_REGION=eu-central-1 \
  -e S3_BUCKET=my-backups \
  -e RESTORE_S3_KEY=pg/mydb/mydb_YYYYMMDDThhmmssZ.sql.gz \
  ghcr.io/kalpak44/postgres-awscli:latest

2. kubectl-awscli

Purpose: a tiny ops image to run Kubernetes + AWS automation inside a Job/CronJob, such as:

  • refreshing ECR pull secrets (regcred)
  • cluster maintenance scripts
  • CI/CD helper tasks inside Kubernetes

Docs: see the latest kubectl-awscli README for:

Quick example (sanity check locally)

bash
docker run --rm ghcr.io/kalpak44/kubectl-awscli:latest aws --version
docker run --rm ghcr.io/kalpak44/kubectl-awscli:latest kubectl version --client

Kubernetes snippet (Job/CronJob-style container)

Use a pinned image tag for repeatable runs, and check the README for the latest recommended pattern.

yaml
containers:
  - name: ops
    image: ghcr.io/kalpak44/kubectl-awscli:latest
    command:
      - /bin/bash
      - -lc
      - |
        set -euo pipefail
        aws --version
        kubectl version --client
        # your script here

Links