Name | Modified | Size |
---|---|---|
.. | ||
minion | ||
p4webapi | ||
pillar | ||
ruby22 | ||
rvm | ||
srv | ||
users | ||
webserver | ||
master | 10 years ago | |
minion-build | 10 years ago | |
minion-buildmaster | 10 years ago | |
minion-dev | 10 years ago | |
minion-eval | 10 years ago | |
minion-master | 10 years ago | |
minion-test | 10 years ago | |
top.sls | 10 years ago |
Change | User | Description | Committed |
---|---|---|---|
13972 | tjuricek | Removing old microservice implementations. The system is now mostly a monolith. Eventua...lly there will be a websocket service. « |
10 years ago |
13788 | tjuricek | Altering the bundler version on the buildmaster and build environments and adding new pill...ars to select the 'run as' user for RVM « | 10 years ago |
13768 | tjuricek | Fixing the version of bundler to 1.8.9. The recent changes to 1.9.x don't seem to work... smoothly with RVM - 'rake install' (a task defined by bundler) seems to shell out to the 'gem' command, which doesn't appear to be configured quite right when launched via an RVM wrapper script. « |
10 years ago |
13707 | tjuricek | Infrastructure for including a "project management" React application. This attempts to... create a fairly simple installer that creates a 'static' folder based on ui/static that gets hosted by the nginx front end. Right now, it's the only app, so the default page is this application. It was called "pws2" during a prototyping phase. Another prototype, "pws" and the related "project" module, is removed since that was a Sinatra-based approach that will be much more difficult to integrate into anything else. I'm running into a couple of issues with notifications setup, it's still not 100%, so I'm disabling this for now from the default 'god' configuration. (The service isn't 100% functional yet, anyway.) « |
10 years ago |
13697 | tjuricek | Add 'unmanaged' node type for configuring a system sans p4d. Not sure about the deploym...ent workflow yet, will likely involve tweaking the Salt config. TBD. « |
10 years ago |
13689 | tjuricek | Switch from Resque to Sidekiq. | 10 years ago |
13679 | tjuricek | Fixed trigger call back to the notifications server by creating a wrapping ruby script. ... Note: also need triggers.io=1 for the server. « |
10 years ago |
13675 | tjuricek | Add notification_services initialization Removing the 'online setup' mode in lieu of do...ing things during the salt process. Mostly this removes the trigger setup from the main web application. « |
10 years ago |
13645 | tjuricek | React.js + Flux prototype This just logs in and displays a list of projects (create one... with the other app for now). « |
10 years ago |
13635 | tjuricek | Exploring modular web applications with Sinatra. If the app has fairly unique content p...er page, Sinatra works really well actually. (Mostly because it's a thin layer over Rack.) Sharing anything though, basically means implementing a module or base class and implementing it everywhere. An added benefit is that these applications could be used in Rails apps as well. I'd probably think hard about this, because Rails tends to be a resource hog. Here's an example config.ru I used for running against the dev system: require 'pms' app = Pms::App.new Pms::Login.settings.p4_web_api_url = http://172.16.100.20/p4_web_api/v1 ProjectApp.settings.phoenix_services_url = http://172.16.100.20/p4_phoenix_services/v1 app.settings.static = :true app.settings.public_folder = File.absolute_path('../../ui/pms', __FILE__) puts "public_folder is #{app.settings.public_folder}" run app « |
10 years ago |
13552 | tjuricek | Added the 'eval' Vagrant box that uses our deployed packages. | 10 years ago |
13544 | tjuricek | Revise package and gem versioning. Packages will use [gem]-[changelist] as their versio...ns. Gems will use a standard Ruby MAJOR.MINOR.REVISION format. P4WEBAPI-64 « |
10 years ago |
13535 | tjuricek | Add notification_services to deployment, and reconfigure build step to exec bash. The e...xecution bit doesn't seem to stay set on config/bash.sh The notification_services service doesn't have advanced tests just yet. « |
10 years ago |
13530 | tjuricek | Add p4_phoenix_services package and Salt configuration for deployment. This uncovered a... couple of issues from the C++ API during it's conversion to C++03. So, in a nutshell, most operations, except for notifications, appear to be working (well, using Vagrant machines). « |
10 years ago |
13528 | tjuricek | Moved rack and app server configuration to be managed via Salt. Also, only using a sing...le value "url" to configure how the p4_project_services instance references the p4_web_api. And, removing the Docker setup, since that won't work for a production system. « |
10 years ago |
13527 | tjuricek | Added a basic p4-project-services .deb package There is some kind of configuration issu...e with the production config that causes tests to fail. The service is running, however, so this is likely related to having a few things not managed via salt. « |
10 years ago |
13526 | tjuricek | Make sure that the p4 project services instance comes up after the libpq-dev package is in...stalled. « | 10 years ago |
13525 | tjuricek | Setup God to manage both the p4 web api and p4 project services processes. Apparently,... there can only be one true god per machine. « |
10 years ago |
13520 | tjuricek | Created a 'cluster' build procedure that creates an installer on build, and executes the i...nstall on a test instance. The main change is to package all gem dependencies via 'vendor/cache' (using the 'bundle package' command). Right now, there appears to be an issue with test data initialization, which may need a revised approach. « |
10 years ago |
13519 | tjuricek | Added a 'buildmaster' environment. The 'buildmaster' sets up Ruby locally for doing som...e work via Rake. Additional steps are now being figured out in the README.md, which may end up being a bash script that executes on the master. « |
10 years ago |
13518 | tjuricek | Revised the 'build' environment to not hard-code the vagrant user, and not assume we have... the p4ws tree available for state.highstate. « | 10 years ago |
13517 | tjuricek | Revised Salt hierarchy to allow for CD clustering. Now, there are two main salt environ...ments: 'build' and 'eval'. The 'eval' environment can be configured for testing or development by setting the Grain 'dev_pillar: True' or 'test_pillar: True'. The test modes may need a bit more effort to figure out exactly where I'll put the .deb files. The dev box passes p4_web_api tests. « |
10 years ago |
13515 | tjuricek | Initial configuration for the p4_project_services. The tests haven't been run yet, so i...t's likely missing some more configuration. Using vagrant to even manage these environments may not be our realistic CD premise. I may end up moving some more responsibility into the Salt layer, which would make vSphere automation easier. « |
10 years ago |
13513 | tjuricek | Update several files that were missing from the last commit regarding a basic 'test' envir...onment « | 10 years ago |
13512 | tjuricek | Add 'test-ubuntu12' environment that sets up projects based on "production" packages. P...ackages are installed from source files that should have been created by the last 'build-ubuntu12' environment. Since the package building process "dirties" up the environment it's better to use a clean system to test package installation. « |
10 years ago |
13504 | tjuricek | The 'build-ubuntu12' seems to generate the appropriate debian packages for the distributio...n. I had permissions issues (probably related to shared folder usage) with the dpkg tools. So I've removed them. If we happen to need them, we'll have to figure out a build process. This environment probably shouldn't be used to test the packages, since omnibus likes to dirty everything up, leave files in the /opt/perforce directory owned by vagrant, etc. « |
10 years ago |
13500 | tjuricek | Reorganizing Vagrant definitions. Each platform should allow developers to launch envir...onments for development, builds, or 'production', so the entire CD process can be evaluated by anyone wiht access to the source tree. This may eventually bring up a transient vSphere machine up and down, which may be necessary for sandboxing Omnibus related builds. « |
10 years ago |
13496 | tjuricek | Complete and seed the setup of the p4webapi in a development mode. This works against t...he local development tree for dev environments. « |
10 years ago |
13495 | tjuricek | Basic p4d configuration for development/evaluation. Uses the perforce-server package an...d just triggers the creation of the helix-evaluation server "instance". This should nicely just ignore the default setup in case these scripts are used in the Helix OVA appliance setup. « |
10 years ago |
13494 | tjuricek | Reorganize salt tree to have a single set of states (probably) organized with different ba...sic minion configurations. « | 10 years ago |
13493 | tjuricek | Attempting to get the p4 web api executing under unicorn via god. The rvm state scripts... deployed as part of SaltStack to not seem particularly reliable. They don't quite seem to execute some of the rvm functions in the correct environments, which means a few basic setup functions are basically shelled out at the moment. « |
10 years ago |
13492 | tjuricek | Some initial salt states for our environent. The ruby setup likely needs some revision.... The goal is to have a system-wide RVM installation set up with gemsets for each app. « |
10 years ago |