Service 03 · Patch management

Updates & security: patch management without taking you offline

In brief

Managed patch management at managed-cluster.net means we apply security patches to the operating system, virtualization layer and services node by node while your systems keep running, critical vulnerabilities first. Exceptions such as a major database upgrade are announced in advance and scheduled in a window you choose. Hardening and versioned configuration are part of the fixed monthly price.

Rolling updates

Managed patch management: rolling updates, node by node

A rolling update patches a cluster one node at a time while the remaining nodes carry the workload. We drain a node using live migration in a Proxmox cluster, update it, verify it and bring it back into the cluster before moving on to the next one. Updates & security is one of four areas of our managed infrastructure operations.

  1. DRAIN THE NODE. Live migration moves its virtual machines to the remaining nodes.
  2. UPDATE. Updates and security patches for the operating system, virtualization layer and services.
  3. VERIFY. The other nodes follow only after this node has been checked.
  4. REJOIN. The node returns to the cluster, and the next node is up.

A faulty update stays on one node

Updates run on a single node first. If verification reveals a problem, it stays contained to that node and we roll the update back.

Diagram 04 Rolling update in a 3-node cluster
  • Live migration of VMs
  • Node active
  • Node being updated
Node 2 has been drained via live migration and is being updated while nodes 1 and 3 carry the virtual machines.
Diagram as text
  1. The virtual machines on node 2 move to nodes 1 and 3 via live migration.
  2. Node 2 goes through four steps: drain, update, verify, rejoin.
  3. If verification reveals a problem, we roll the update back.
  4. Then the next node follows; only one node is in maintenance at any time.

Exceptions

Updates without maintenance windows, and the exceptions

Rolling updates without downtime and without maintenance windows are our standard: we apply updates node by node while your virtual machines keep running on the other nodes. Exceptions such as a major database upgrade are announced in advance and scheduled in a window you choose, because such an upgrade may require a brief switchover.

Who does what Updates at a glance: responsibility, method and timing
Type of changeResponsibleMethod and timing
Updates and security patches for the operating system, virtualization layer and servicesWeRolling update node by node while the cluster keeps running, with live migration of the virtual machines beforehand.
Critical vulnerabilityWePrioritized, as a rolling update. Until a patch is available, interim measures such as firewall rules where possible.
Major database upgradeWe Scheduling: youMay require a brief switchover. We announce the upgrade and carry it out in the window you choose.
Updates to your application, code and extensionsYouYou take care of your application, we take care of everything underneath it. Our monitoring uses health checks from the user's perspective to verify that it responds.

You have a dedicated contact who knows your cluster.

Vulnerabilities

Security patching service: critical vulnerabilities take priority

We handle critical security vulnerabilities in your cluster's operating system, virtualization layer and services with priority. The patch is applied as a rolling update, node by node, so the cluster keeps running. Until a patch is available, we put interim measures such as firewall rules in place where possible.

Vulnerabilities in your application, your code and your extensions remain your responsibility, since you maintain the application. Our security patches cover the platform underneath it.

An overview of our measures for secure operations is on our security and compliance page: security measures at a glance.

Patch deadlines from your security policy

If your policy sets deadlines per severity level, raise it with us. We will tell you clearly what we can demonstrate today.

Hardening

Server hardening: a smaller attack surface on every system

Server hardening reduces a system to what operations require and secures the access that remains. It complements patch management: patches close known vulnerabilities, while hardening shrinks the attack surface. Operations at managed-cluster.net include system hardening built on four measures: minimal services, restrictive firewalls, secured SSH access and network segmentation.

Minimal services

Systems run only the services that operations require. This reduces the attack surface and the number of components that need updates.

Restrictive firewalls

Firewalling at every layer, plus a redundant firewall and load balancer tier with health checks.

Secured SSH access

SSH access to the systems is secured. The Enterprise package adds separate networks, VPN, IP whitelisting and role-based access.

Network segmentation

Management, storage and public traffic run on separate networks (VLANs); storage has its own dedicated network.

On top of this, network-level DDoS protection, included in every fixed price, helps defend against denial-of-service attacks from the internet.

Configuration

Versioned configuration and TLS management

At managed-cluster.net, updates and security include maintaining your configuration: we keep it versioned and traceable, so every change can be followed and earlier versions are retained. We also look after TLS certificates, renewing them automatically and monitoring them so that no service fails because of an expired certificate.

Documentation of the architecture, access details and emergency plans is provided in writing and belongs to you.

Backup

Backup strategy: stored separately from the cluster

Our backup strategy keeps backups apart from the cluster: they run on separate infrastructure in German data centers, with defined retention periods and regular restore tests. In the Business package they are stored, and restore-tested, at the second site NBG01. Our page on backup strategy and restore testing covers backup, retention and recovery in detail.

A high availability cluster is not a backup.

Replication protects against hardware failure, yet it also copies errors and accidental deletions to every replica.

Security incident

Security incidents: analysis, containment, recovery, post-incident report

When a security incident occurs, we follow a defined four-step process: analysis, containment, recovery and a post-incident report. Critical alerts reach on-call duty immediately and are handled right away, around the clock. How we monitor and alert is described on our page on 24/7 monitoring with on-call duty.

  1. 1

    Analysis

    We determine which systems are affected and how the incident happened.

  2. 2

    Containment

    We contain the incident so that it does not spread to other systems.

  3. 3

    Recovery

    We bring the affected services back to a secure state, restoring from the separate backup if needed.

  4. 4

    Post-incident report

    You receive a post-incident report covering the root cause and the measures derived from it.

FAQ

Frequently asked questions about updates and security

How are updates carried out?

As rolling updates, node by node, while the cluster keeps running. We drain a node via live migration, apply the updates, verify the node and bring it back into the cluster; only then do the remaining nodes follow.

Are there maintenance windows?

Only for exceptions. We apply updates and security patches node by node while your systems keep running, without maintenance windows for your users. A major database upgrade may require a brief switchover; such exceptions are announced in advance and scheduled in a window you choose.

How do you handle critical security vulnerabilities?

With priority. The patch is applied as a rolling update, node by node. Until a patch is available, we put interim measures such as firewall rules in place where possible. If your security policy sets deadlines per severity level, raise it with us.

What happens if an update causes problems?

We roll it back. Updates run on a single node first and are verified before the remaining nodes follow, so a problem stays contained to that node.

What do you do in a security incident?

We analyze it, contain it, restore the affected services and write a post-incident report covering the root cause and the derived measures. Critical alerts reach on-call duty immediately, around the clock.

Next step

Does our patching process fit your security requirements?

Tell us your requirements for patch management, hardening and maintenance windows. We will walk you through the process for your cluster and the evidence you receive during operations: a post-incident report for every incident, a monthly operations report and documentation of your systems.

Discuss patching