Open tobiasschweizer opened 7 years ago
here are a handful of suggestions of how to avoid direct configuration file editing without using puppet or another configuration management system.
by mapping directories to mount points in the docker instance, and proxying local ports to hosted ports in the docker instance.
As it is all web-services, they are easy to proxy.
heart of a release process:
git stash
git pull
git stash apply
I described earlier the set-up we use here, in a now non-readable form (because I didn't use plain text mail): https://www.maillist.unibas.ch/pipermail/knora-user/2016-July/000079.html
it is a mix of apache web-proxy and git-stash for release (but we plan on using docker).
When setting up Knora on a server that has an own domain (non localhost), a bunch of (partially redundant) modifications has to be made. Some of them are URLs with or without the port, some of them are just a domain name (without
http://
or port):webapi/src/main/resources/application.conf
:app.http.interface
: Knora API URLapp.http.port
: Knora API portapp.http.base-salsah-url
Salsah webapp URL + portapp.tmp-datadir
app.datadir
app.sipi.url
: Sipi URLapp.sipi.port
: Sipi portsalsah/src/main/resources/application.conf
:app.http.base-knora-url
: Knora API URL + portsalsah/src/main/scala/org/knora/salsah/Main.scala
:salsah/src/public/js/00_init_javascript.js
:Can we simplify that and make it more consistent?