strong-pm

strong-pm

Quick Start

Install and run strong-pm server:

npm install -g strong-pm
sl-pm -l 7777

Clone and push an app:

git clone [e-mail]:strongloop/sls-sample-app.git
cd sls-sample-app
git push http://localhost:7777/repo HEAD

That was a non-production push, it installed all your dependencies on the server. You should always built your app so the dependencies are built-in, and not installed dynamically at run-time:

slb --install --commit
git push http://localhost:7777/repo HEAD

Config notes

Config is an .ini file. Configurable are:

• prepare command: defaults to npm rebuild; npm install --production
• start command: defaults to sl-run XXX should default be clustered?
• stop signal: defaults to SIGTERM

Configuration can be global, or per repo pushed to.

Push from git with a command like git push http://localhost:PORT/REPO

PORT is arg to --listen, REPO you decide but is not optional

Example:

; these are the defaults:
prepare[] = npm rebuild
prepare[] = npm install --production
start = sl-run
stop = SIGTERM

; these are overrides for a particular repo
[config-one]
; no prepare
prepare =
start = node .
stop = SIGHUP

Architecture:

    ------
    node
      cluster | net | ...
    ------------------
    strong-   |
    cluster-  |
    control   |
    ------------------------
    strong-
    supervisor
       uses: fs, syslog, run-time control, strong-agent, s-c-c, signals, ...
    
       XXX the control channel/pipe should not be in cluster-control!
    ---------------------------------
        upstart      |     strong-pm    
    how to restart   |   git receives, and SIGHUPs to trigger chdir/and
    becomes your     |   worker restart
    problem, but not |
    a problem if     |
    you deploy in
    containers