Hello Eugen and all, This is really a great thing to do. I am sure a lot of devs will be interested in this, Will be happy to collaborate.
Regards, Raja On Thu, Jul 1, 2021 at 8:59 PM Carsten Schinzer < cars...@dcs-verkaufssysteme.de> wrote: > Hello Eugen and all, > > > I like the initiative and I am convinced this is a forward-looking > approach. > > As you have read in the comment from Pierre the hurdle to overcome is the > external dependency injection of e.g. database connectivity as > configuration maps at runtime. In my company in Kubernetes stacks we are > solving that by maintaining a config map of RDBMS connectivity per > application stage and region and deploy that. The app then reads the config > information at container startup. > > When taking the regular OFBiz approach, though, you would need to know at > build time to which stage (Staging, production, Development) and which > region you are going to deploy, which will be a bit of an overkill if you > want to leverage the full flexibility of a cloud-based deployment which > typically comes with variance on DB connection strings. > > Alternatively, you would need to unpackage certain release artifacts and > rewrite them as part of the deployment process, which might be an > antipattern on Helm. > > Could that be overcome and dynamic DB connection string be used somehow? > Warm regards > > > Carsten > > --- > > > Dr. Carsten Schinzer > > *Inhaber* > > > t +49 89 88569642 | f +49 89 99964059 | m +49 159 05269462 > > DCS Verkaufssysteme > > Gerner Str. 27 | 80638 München | Germany > > > Am Do., 1. Juli 2021 um 10:57 Uhr schrieb Pierre Smits < > pierre.sm...@gmail.com>: > > > Hi Ioan, > > > > My apologies for the late reaction. > > > > IMO we can forget about the Derby route for this. Derby is not production > > grade, and often adopters/implementers already have their preferred RDBMS > > for production in play or decided upon. Configuration of such is the > > entityengine.xml file, and deployment of such for the various > > environments/instances like UAT or PROD can easily managed and deployed > via > > VCS. > > > > Best regards, > > Pierre Smits > > > > > > On Sun, May 9, 2021 at 9:13 PM Eugen Stan <eugen.s...@netdava.com> > wrote: > > > > > > > > > > > Hello, > > > > > > NOTE: I previously sent this email from an un-subscribed email address, > > > please ignore that one. > > > > > > I'm working on a Helm chart for OFBiz > > > https://github.com/ieugen/charts/tree/main/ofbiz . > > > > > > The goal is to have OFBiz running in Kubernetes. > > > > > > Right now it's a rough start: it starts OFBiz with Derby and that's > > > about it. > > > > > > I'll keep pushing updates as soon as I have them. > > > > > > I'm interested in your feedback. > > > If you plan on testing it and using it let me know. > > > I do accept and encourage collaboration. > > > > > > If this is interesting I might push it upstream but it uses the binary > > > build of OFBiz in Docker. > > > > > > > > > I'm using docker images built from trunk on jdk-11. > > > There are amd64 and arm64 (Raspberry PI4) images. > > > > > > Some issues that need solving are the many configuration files that > > > OFBiz has. > > > > > > This details creates a lot of configuration when deploying in Docker / > > > Kubernetes as they need to be mounted in a volume with many entries or > > > multiple volumes. > > > > > > There are many potential solutions to this but it boils down to: > > > * It would be nice to load max 1-2 or 3 configuration file with > multiple > > > sections instead of the currently many configs split over the code > base. > > > * It would be nice to support configuration (overwrite) via ENV vars > for > > > some things like DB connection. > > > > > > > > > Regards. > > > Eugen > > > > > > -- Thanks, Raja
