I am looking online for documentation that pertains to running Narayana on Kubernetes (not openshift). I have found several resources online, but have really not found adequate material on this critical question (critical in my mind due to the increasing focus on kubernetes enabled services).

I am creating this issue to get help from the experts, and perhaps to help create some further energy and support behind this need.

Resources that I have found that seem to relate include.

• https://narayana.io/
• https://github.com/jbosstm
• https://github.com/jbosstm/quickstart/tree/master/jts-docker
• https://access.redhat.com/documentation/en-us/red_hat_fuse/7.2/html/fuse_on_openshift_guide/camel-spring-boot_xa
• https://github.com/jboss-fuse/spring-boot-camel-xa
• https://blog.joshdreagan.com/2017/08/14/transactions_and_alternatives_with_camel/
• https://github.com/joshdreagan/camel-spring-boot-xa

I am currently setting up a prototype system, and will carefully document the configuration steps and configuration files required. I'd be glad to capture some of that detail on this issue, to support the narayana team in capturing documentation.

My current appraoch is

1. Run Narayana on k8s (rather than standalone).
2. Use an external postgres object store (also running on k8s).
3. Have the postgres object store use Ceph-RBD based PV’s (in other words ensure that their storage is distinct and separate from the processing nodes both for JTA and for Postgres)

Am I crazy to want to run narayana on k8s? Can it be done in a fashion that ensures proper safety of the transactions that it manages? I believe the answer is no to the first question and yes to the second.

Suggestions/Thoughts/Guidance from the experts would be appreciated.