Vetted engineers with verified commercial track record in Server Administration
Gliwice, Silesian Voivodeship, Poland (Gliwice, Poland)
IT Systems Administrator and self-employed infrastructure specialist (Linux | Windows Server | VMware | Cisco | Zabbix)
Independent Polish system administrator and infrastructure specialist. Provides audit, deployment, and support services for Linux and Windows Server servers, MSSQL database configuration, VMware ESXi virtualization, Cisco and FortiGate networking solutions. Handles regular backups (NAS/FreeNAS), implements Zabbix monitoring systems, and automates daily tasks with Python.
London, United Kingdom (London, United Kingdom)
Senior Research Infrastructure Engineer at Imperial College London | Linux | HPC | Ansible | Slurm
Lead server and HPC infrastructure engineer with 10+ years of experience in Linux systems administration, automation, network and storage configuration. Manages a large-scale Linux server platform at Imperial College London: RHEL 8/9 administration on bare-metal servers and VMware virtualization (clusters with 1,000+ nodes), maintenance of GPFS / IBM Spectrum Scale parallel filesystems (500+ systems), Mellanox InfiniBand / RDMA networking, Slurm and PBS workload schedulers. Automates configurations with Ansible, Terraform, and GitLab CI/CD, provides monitoring via Grafana and InfluxDB.
Rottenburg / Tübingen, Baden-Württemberg, Germany (Germany)
Linux System Administrator | HPC Cluster Administrator at Tübingen AI Center
Linux infrastructure and cluster system administrator. Specializes in designing, deploying, and maintaining virtual environments on Proxmox VE, building fault-tolerant Ceph storage, automating configurations via Ansible and AWX, and managing infrastructure as code (Terraform). Ensures stability of bare-metal server nodes, network equipment (VLAN, pfSense, MikroTik), and monitoring systems.
Server administration is the work of configuring and operating the systems that host your application. VPS management focuses on your virtual private server: access, services, resource usage, updates, backups and the instructions needed to maintain it.
A freelance sysadmin can help with a single problem or an agreed maintenance scope. State which operating system and hosting provider you use, what runs on the server and whether customers currently depend on it. The administrator can then assess the work and the access required.
For a new VPS, define which application and database it will run, who needs access and how updates will be delivered. Request a documented configuration and a demonstration of the main application workflow after deployment.
For ongoing support, agree what is monitored, how alerts reach the responsible person, when maintenance happens and how backups are checked. A support arrangement should state response windows and exclusions; a listed profile does not establish round-the-clock coverage.
A server can be reachable while the website it hosts is failing. A useful diagnostic brief includes the affected URL, the observed error, the time it started and recent changes. Logs and service checks help distinguish a proxy issue from an application or database failure.
Searches such as custom 502 page nginx and custom error page nginx describe a presentation requirement. A custom error page can explain an outage to visitors, but the underlying service failure needs its own investigation. Include both the desired error-page behaviour and the reliability problem in your brief.
Ask for a backup scope that identifies the data to preserve, where copies are stored and how a restore will be checked. A recovery plan should name dependencies and describe who can perform it when the original administrator is unavailable.
For a migration, list the applications, databases, scheduled jobs, domains and external services involved. Agree how the new environment will be validated, when traffic can move and what conditions require a rollback. Downtime expectations depend on the actual setup.
Hire a sysadmin when the main task is server operation: configuration, maintenance, access, backups or troubleshooting. A brief to hire a DevOps developer often adds deployment automation and coordination between application changes and infrastructure. The roles overlap, so select by the work and evidence rather than the title alone.
Describe your problem without sharing passwords or private keys in the public project form. After agreeing the scope, arrange the appropriate access privately and retain ownership of the hosting account and domain.
A single Ubuntu 24.04 VPS with Nginx 1.26, UFW, Fail2ban, and unattended upgrades fits one app with steady traffic and a small team. A HestiaCP or CloudPanel host adds click-to-manage sites at the cost of update coupling. Docker 26 with Compose isolates app, PostgreSQL 16, and Redis 7 for repeatable deploys. A move to Hetzner or DigitalOcean fits expiring contracts or chronic noisy-neighbor load. Independent sysadmins run $45-$95/hr against $150-$250/hr at agencies, with matching in 2-4 hours.
Expect 4-10 hours for a hardened single VPS with backups, 6-14 hours for a panel setup with per-site PHP or Node versions, 12-28 hours for containerization with restic backups and Uptime Kuma checks, and 10-24 hours for a host migration with DNS cutover in Cloudflare. Pick single VPS for simplicity, panel for non-technical owners, containers for multi-service apps, and migration when monitoring proves the host is the fault.
| Option | Cost / time | Limits | Pick when |
|---|---|---|---|
| Single VPS hardened | $45-$95/hr, 4-10 hrs | One failure domain | One app, steady traffic, small team |
| Panel host | $45-$95/hr, 6-14 hrs | Panel update coupling | Owner needs click management |
| Docker Compose | $55-$95/hr, 12-28 hrs | Needs image discipline | App plus database plus cache |
| New host migration | $50-$95/hr, 10-24 hrs | DNS propagation wait | Contract or hardware faults persist |
Takeovers often show root-only SSH, three Nginx conf.d copies, expired Certbot cron jobs, a 400 GB log partition, and MySQL 5.7 dumps nobody restores. Phase one (2-5 hours) records running services with systemctl, open ports with ss, cron tables, and disk use with ncdu, plus a restic snapshot stored off-host. Nothing is deleted before the snapshot restore is proven in a staging droplet.
Phase two (8-20 hours) rebuilds one service per window: least-privilege users, UFW rules, Nginx with tested 502 pages, TLS renewal dry-runs, and Prometheus node exporter plus log rotation. Phase three (4-10 hours) rehearses cutover with lowered Cloudflare TTL, database import timing, and a rollback DNS flip. Traffic moves only after the new host serves error-free logs for 24 hours. Keep screenshots of each passing health check.
Migration guardrail
Lower DNS TTL to 300 seconds a day before cutover. Skipping this turns a ten-minute rollback into a multi-hour outage.
A takeover starts with read-only facts: hosting login owner, Ubuntu or Debian version, Nginx configs, systemctl status, Certbot lineage, backup locations, and the last outage timeline. A sysadmin spends 2-4 hours checking SSH key sprawl, unattended-upgrade logs, disk trends, and whether backups in Backblaze B2 or S3 actually restore. You receive a keep-versus-rebuild verdict with hour bands before changes begin.
The audit separates cosmetic debt from danger: world-writable keys, password auth still on, no off-host copies, and 502 errors caused by a dead upstream rather than Nginx itself. Cleanup scopes of 8-20 hours fit hosts with salvageable configs. When every service runs as root with no snapshots and unknown cron writers, a fresh VPS build costs less than forensic repair.
| Signal in audit | Meaning | Response |
|---|---|---|
| Root-only SSH, password auth | Compromise-prone access | Key-only rebuild, 3-6 hrs |
| No off-host restore test | Backup is fiction | restic setup plus drill, 4-8 hrs |
| 502 with dead upstream | App fault, not proxy | Fix service units, 2-6 hrs |
| Unknown crons, full disk | Unowned automation | Fresh VPS cheaper than repair |
Rebuild threshold
If access cleanup, backup repair, and config rebuild exceed 60% of a fresh VPS estimate, approve a clean Ubuntu 24.04 build and migrate data with checks.
Screen for Ubuntu 22.04/24.04 fleets with UFW, Fail2ban, Nginx reverse proxies, Certbot renewals, and restic or Borg restores the candidate has actually performed. Ask how they diagnose a 502 across Nginx error.log, journalctl, and upstream health, how they size_swap and tmpfs for a small VPS, and what their backup test cadence is. Seniors cite log lines, timer units, and restore minutes. Juniors describe dashboards without shell commands or rollback detail.
Run a paid 3-5-hour trial on a staging droplet: harden SSH to key-only, configure Nginx with a custom 502 page and HTTPS, and prove a restic backup and restore of one database. Acceptance means a written runbook, green Certbot dry-run, Uptime Kuma check passing, and hours logged per command batch. Matching takes 2-4 hours; independents bill $45-$95/hr versus $150-$250/hr at agencies.
Common terms are VPS administration, VPS management and VPS support. For Linux systems, Linux server administration is more specific. A person doing the work is usually called a system administrator or sysadmin.
You can define a one-off scope such as server setup, a deployment issue, an Nginx error or a migration. Agree the acceptance criteria and any follow-up support with the specialist.
Only if that is explicitly agreed. Define monitoring coverage, who responds to alerts, response windows and the maintenance schedule before starting a support arrangement.
Yes. An on-call system administrator can immediately inspect systemd process supervisors, Nginx error logs, upstream application processes, and memory exhaustion to restore production availability.
The estimate depends on the number of servers, applications, urgency and support coverage. Separate initial assessment, implementation, hosting costs and ongoing maintenance in the proposal.
Choose a hardened single VPS for one steady app (4-10 hours), a panel for click management, Docker Compose for multi-service isolation, or a new host when monitoring proves hardware faults. A 2-4-hour audit at $45-$95/hr settles the pick.
2-5 hours for an audit and snapshot, 4-10 hours for a hardened VPS, 12-28 hours for containerization, and 10-24 hours for a migration. Emergency 502 triage usually starts with a 1-2-hour diagnostic before a fix estimate.
Yes, with hosting ownership, OS version, Nginx configs, and backup locations. A 2-4-hour audit checks access, restores, and failure causes, then gives a keep-versus-rebuild verdict with hour bands.
Give a paid 3-5-hour staging task: key-only SSH, Nginx with custom 502 page and HTTPS, plus a proven restic restore. Pass means runbook, green Certbot dry-run, passing Uptime Kuma check, and logged hours.
Provide hosting account ownership, domain registrar path, OS and app versions, current Nginx configs, and the last outage timeline. Arrange SSH keys privately after scope sign-off; never post passwords or private keys.
Agree 2-5 hours per month at $45-$95/hr for patches, backup-restore drills, cert renewals, and log review, with named response windows for outages. Keep the old VPS 7 days after migration as fallback.
