The Unholy Union of Corosync, Pacemaker and CTDB


One Line Summary

Discuss integrating CTDB with Pacemaker clusters; we need one less messaging layer, and one less resource manager.


Corosync plus Pacemaker provides a mature, generic HA clustering solution, but cannot by itself run clustered Samba.

CTDB gives us clustered Samba, but also has its own messaging layer and event scripts for resource management.

CTDB on Corosync+Pacemaker works, but then we have two messaging layers, and two possible places to manage highly available resources. This is confusing to manage, and it’s unfortunate that the same set of problems is being solved twice.

The goal here is to discuss ways the two projects can integrate more effectively, to reduce development effort and ease system administration.


ha, clustering, corosync, pacemaker, ctdb


  • Tim Serong



    Tim is currently employed by Novell as Senior Clustering Engineer, working on the SUSE Linux Enterprise High Availability Extension. He has been building, breaking and fixing Pacemaker clusters for the past several years, through which experience he has discovered that you can never ever have too many log files.

    Interests include motorcycling, beer, and making technology simpler and more reliable (although not necessarily in that order, and usually not all at the same time).

Leave a private comment to organizers about this proposal