SCM Manager High availability

Is it possible to install SCM Manager as a High Availability solution? If so how we can achieve this?

Hello Tomas,

at the moment it is not possible. We are aware of this issue, but there are still too many technical obstacles. For example, we have no idea how to handle Mercurial or Subversion in such a highly available environment. Also, we increase performance by keeping data in-memory, which would no longer be possible.

It feels like we need to rewrite SCM-Manager and create a version 3.x or something. I wouldn’t rule out the possibility of us addressing this issue at some point. But currently it’s not on our roadmap and we don’t have the time / budget for it.

Kind regards, Eduard

I have achieved the HA with the help of DRBD, Pacemaker, Corosync, Keepalive, and HA-Proxy combination on the Rocky Linux cluster. I have tweaked the default configuration of the SCM-Manager to have all the memory components in one DISK and DRBD will take care of making it available on the other node. The Pacemaker will take care of switching the complete solution if any node goes bad. This is a true HA-Cluster working for me for my Jenkins for over 8 years. I have just ported the same for SCM-Manager and customized it.

It is very well documented with clear working commands.
I can have it documented here if it is useful.

1 Like

Hey @girishkg

that sounds great! I really would like to know how you made high availability for SCM-Manager work!

Thank you for sharing your knowledge with the SCM-Manager community. Really appreciated :+1:

1 Like

Hi @christoph.loose ,
I would like to document it here on the cloudogu,

  1. How do you want me to start?
  2. Any specific page or the tree?
  3. Can we have a call if required?

Sure, let’s have a call for a kick off.

Could you please send me your availability? christoph.loose@cloudogu.com
Then I can organize a meeting in Google Meet