Building a Nextcloud Cluster A word about.. do you really need a Cluster?

image_pdfimage_print

As we all know, High Availibility Clusters, cause as much Problems, as they solve Problems..

This is the only truth, i head to learn over the years, it doesn’t matter which Cluster Framework we talk about…

Corosync/Pacemaker
Solaris Cluster
Veritas Cluster

The Problems are legion.
High availibility Cluster don’t improve Performance, like some Decision Makers think.

In most Cases it’s questionable if you really need High availibility Framework.

If you need a High Availibility Framework. If you’re a decision Maker!

Make damn sure that your Technicians are well educated, and have a deep understanding about Cluster Concepts!

It’s not a real Problem to Copy and Paste from a How To… in 95% of all Cases the stuff will work… til

D-Day = Desaster Day will happen. And you can be sure Murphy will strike you ! Thats the only rule we all have to face.

On this particular Day, better make sure your Operators are well educated, to avoid the Desaster.

Ask yourself if you’re not better suited with a Cold Standby Option, and a well working Backup Solution.

Also consider the Option if you’re heavyily using Virtualision Products, like Vmware Vsphere, Nutanix, Hyper V oder plain KVM, to leave the High Availibilty Topic to your Virtualisation Product.

If you’re coming to the Conclusion you need a Linux High Availibility Cluster, consider that you better decide for one Linux Distribution..

Why do i tell you this?

As often in FOSS the several Vendors, are not able to find a practical agreement.

SuSE Ubuntu/Debian and the majority of the Linux World use the crm Shell Interface for configuring, implementing and managing the HA Cluster Stack.

Redhat decided to reinvent the Wheel again, one day they came up with the
PCS Shell.

I’ll assure you, under the hood, the Cluster Framework, is nearly the same on all Linux Distributions. They only use different versions. But the Corosyn Pacemaker stuff is the engine of all Linux Distributions, who provide a High Availibility Solution.

I have a very strong opinion about how Redhat’s move, in first case it is complete Horseshit to reimplement a working Solution again, for the Same Product, that every other Distribution is using. Seems to be the „Not invtended here Syndrome“.

Secondly it’s a kind of Open Source Vendor Lockin.

If you have several Platforms running, your Operators have to be skilled with the CRM Shell and the PCS Shell.

SuSE is offerin a crm Shell packaged for CentOS7 and RHEL7
But make yourself aware, if you make use of the CRM Shell on RHEL7 and you have a Support Contract with Redhat it might be possible that you run a unsupported Setup, think about it, before you open a ticket.

After reading this, you may ask yourself, why is this Guy implementing the Nextcloud Application with an HA Cluster?

First.. i’m able to soy, it’s that simple.

Secondly i’m making a living out of building and supporting High Availibilty Clusters.

Question answered? 🙂

If you’re curios about how to implement the Nextcloud in the Cluster follow the next Chapters 🙂

Building a Nextcloud Cluster A word about.. do you really need a Cluster?