Deployment evolution to releases

[Darren S.]

Hi,

I wanted to find out what the project's plans are, if any, for
advancing the deployment of MISP in the future.

The current deployment model is fairly involved and while not terribly
problematic for one-off deployments on single hosts, can be a bit
cumbersome when deploying multiple instances or when attempting to
roll the deployment into a software config management system. We have
been working for some time on a flexible Ansible deployment role for
MISP 2.4 for example, and have found it to be problematic in various
ways. We also consider the possibility of creating distribution
packages (e.g. DEB packages, RPMs, BSD ports, etc.) and see
complications there.

Most PHP apps are deployable by unpacking an archive (tarball, etc.)
into the web server document root and running the necessary SQL
commands to create/populate database schema, and then modifying the
config file(s). This is true for many applications, although it should
be acknowledged that many applications are much simpler in
design/purpose than MISP.

In my experience, the parts of current MISP deployment that provide
the most challenges involve the way that the current deployment
method, as well as the update channel and population of certain
built-in dependencies, are sourced directly from Git as version
control. Version control based deployments have become more popular in
recent times but in addition to some conveniences, also force some
complications on the administrators. We seem to have continual
problems attempting to update between versions within the 2.4 codebase
using the git submodule methods, falling into various complications
related to changes between versions, modified files in the tree, and
general lack of complete familiarity with Git - which arguably,
shouldn't be a requirement for the user of software, but rather the
developers alone.

If the project were to build releases of the software, it would allow
the PHP application to more closely resemble similar applications and
deployment and updates/upgrades could potentially be eased for end
users and open up possibilities for carrying MISP into distribution
package repos. Just curious the project's appetite for this.

--
Darren Spruell
phatb...@gmail.com