#dokku

/

      • savant1 joined the channel
      • savant has quit
      • savant1 has quit
      • c0b has quit
      • savant joined the channel
      • savant has quit
      • savant joined the channel
      • savant has quit
      • ome joined the channel
      • savant joined the channel
      • cdbattags joined the channel
      • phrozen77 has quit
      • jalcine has quit
      • Friss joined the channel
      • FMJaguar_ has quit
      • FMJaguar_ joined the channel
      • phrozen77 joined the channel
      • humerusj joined the channel
      • jalcine joined the channel
      • benwilber joined the channel
      • shykes joined the channel
      • abl joined the channel
      • tanepiper joined the channel
      • kudos joined the channel
      • andrewstewart has quit
      • andrewstewart joined the channel
      • davidj joined the channel
      • bongjovi joined the channel
      • savant has quit
      • raydeo_ joined the channel
      • bongjovi has quit
      • savant joined the channel
      • savant has quit
      • bongjovi joined the channel
      • p9cl joined the channel
      • p9cl
        hello. what's the best way to map in a bunch of tcp ports (not http). my best guess at this point is to use docker-options and -p but i understand that may lead to problems when deploying/restarting?
      • there seems to be another way to do it using nginx and proxy:ports, but i can't find an example that isn't http.
      • zoka joined the channel
      • zoka has quit
      • zoka joined the channel
      • zoka has quit
      • phrozen77 has quit
      • Friss has quit
      • oxsyn has quit
      • gliderlabs has quit
      • sung has quit
      • shingara has quit
      • mgolisch has quit
      • forgot has quit
      • oxsyn joined the channel
      • phrozen77 joined the channel
      • Friss joined the channel
      • gliderlabs joined the channel
      • shingara joined the channel
      • mgolisch joined the channel
      • forgot joined the channel
      • sung joined the channel
      • savant joined the channel
      • savant
        p9cl: you'll need to do the following:
      • - disable the proxy plugin for your application
      • - disable zero-downtime deploys
      • - use the docker-options plugin with the `-p` flag
      • thats the current way to do it
      • unless you'd like to add support for tcp proxying to either our core nginx-vhosts plugin or write your own
      • p9cl
        okay, thanks. -p it is then.
      • savant has quit
      • tref joined the channel
      • c0b joined the channel
      • cdbattags joined the channel
      • [d__d] joined the channel
      • cdbattags has quit
      • gliderlabs
        [Steve Hobbs] Hi again, I just had a quick question. I noticed there that when I redeploy my dokku app, it starts a new container instead of replacing the existing one. I usually have to go in and stop/delete the old running container manually - is that intended?
      • tx
        the old container usually stops after a while
      • it brings the new one up before taking the old one down as a check
      • gliderlabs
        [michaelshobbs] the default wait time to retire an old container is 60 seconds. this can be overridden by setting the DOKKU_WAIT_TO_RETIRE config var
      • [Steve Hobbs] Ah I see, thanks. I did just check it immediately after deployment right enough
      • Duller999 joined the channel
      • Duller999 has quit
      • savant joined the channel
      • bongjovi has quit
      • bongjovi joined the channel
      • bongjovi has quit
      • c0b has quit
      • cdbattags joined the channel
      • cdbattags has quit
      • [Reggie Chen] Hi, I am testing a simple node js app from https://github.com/heroku/node-js-sample with my own base container that I created to replace gliderlabs/herokuish . It is just a simple container based on cedar:14 and add MONO dev stack onto it. This container, together with Dokku 0.4.3 help me to automate building our node/.net mix application for quiet sometimes since last year. On version 0.4.3 , I was able to verify
      • [Jose Diaz-Gonzalez] if you execute a binary from a direct path, it should be there
      • [Jose Diaz-Gonzalez] dokku doesn't change that, it simply executes your image
      • [Jose Diaz-Gonzalez] are you sure its still using your same base container?
      • savant has quit
      • savant joined the channel
      • [Reggie Chen] I think so. What I did is I tar my image that is working with my build server running in digital ocean. Download the tar file, and load it with 'docker load -i..', , and build the test application with version 0.7. Before initiate 'git push..', I make sure I've exported DOKKU_IMAGE with the custom image tag, not the default gliderlabs/herokuish...
      • [Jose Diaz-Gonzalez] how are you exporting it?
      • [Reggie Chen] You can checkout my Dockerfile at https://github.com/lazydino/herokuish/blob/magn...
      • [Reggie Chen] I do it like 'export DOKKU_IMAGE=magnetar/heroku:mono' ?
      • [Jose Diaz-Gonzalez] on the host?
      • [Jose Diaz-Gonzalez] or where?
      • [Reggie Chen] Yes, on my local machine. not remote reference.
      • [Jose Diaz-Gonzalez] yeah that wont affect the remote server
      • [Jose Diaz-Gonzalez] if you want an app to use your custom image
      • [Jose Diaz-Gonzalez] you need to do something like
      • [Reggie Chen] yes, and 'git push..' is pushing to my localhost server too
      • [Jose Diaz-Gonzalez] you just said on your local machine, not on the remote
      • [Jose Diaz-Gonzalez] `dokku config:set --no-restart APP DOKKU_IMAGE=magnetar/heroku:mono`
      • [Jose Diaz-Gonzalez] then it will use your custom image for that app
      • [Reggie Chen] OK, I will try.
      • [Reggie Chen] Sorry about the confusion, the testing is done all locally, the custom image is store in local too.
      • CanTonic joined the channel
      • savant has quit