bengler / brow Goto Github PK

View Code? Open in Web Editor NEW

4.0 4.0 1.0 265 KB

A tool that configures unicorn and nginx locally for pebbles development

Ruby 71.70% HTML 28.30%

brow's People

Contributors

Stargazers

Watchers

Forkers

atombender

brow's Issues

brow start <someproject> doesn't start nginx

When nothing is running, then starting a single project, e.g. snitch with

$ brow start snitch

fails to start nginx.

This results in the following:

$ curl 'http://snitch.dev/api/snitch/v1/ping'

<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
<html><head>
<title>404 Not Found</title>
</head><body>
<h1>Not Found</h1>
<p>The requested URL /api/snitch/v1/ping was not found on this server.</p>
<hr>
<address>Apache/2.2.21 (Unix) DAV/2 Server at snitch.dev Port 80</address>
</body></html>

Brow watch fails during startup

This is what happens:

$ brow watch

==> /tmp/brow-checkpoint.stderr.log <==

==> /tmp/brow-connector.stderr.log <==

==> /tmp/brow-dittforslag.stderr.log <==

==> /tmp/brow-grove.stderr.log <==

==> /tmp/brow-kudu.stderr.log <==
/Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/guard-0.9.1/lib/guard/listener.rb:36:in `[]': can't convert Symbol into Integer (TypeError)
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/guard-0.9.1/lib/guard/listener.rb:36:in `select_and_init'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/brow-0.0.1/lib/brow/watcher.rb:46:in `watch'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/brow-0.0.1/lib/brow/watcher.rb:21:in `block in start'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/brow-0.0.1/lib/brow/watcher.rb:21:in `each'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/brow-0.0.1/lib/brow/watcher.rb:21:in `start'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/brow-0.0.1/lib/brow/wrangler.rb:81:in `watch'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/brow-0.0.1/bin/brow:74:in `watch'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/thor-0.14.6/lib/thor/task.rb:22:in `run'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/thor-0.14.6/lib/thor/invocation.rb:118:in `invoke_task'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/thor-0.14.6/lib/thor.rb:263:in `dispatch'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/thor-0.14.6/lib/thor/base.rb:389:in `start'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/gems/brow-0.0.1/bin/brow:89:in `<top (required)>'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/bin/brow:19:in `load'
    from /Users/bjoerge/.rvm/gems/ruby-1.9.2-p290/bin/brow:19:in `<main>'

Syslog on Mac does not log anything below WARNING, so there is no logging [housekeeping branch]

So how do we get logging from Unicorn etc.?

Needs to work on Ubuntu (and possibly other linuces)

Brow currently works only for OS X. These are the parts that are known to be platform specific at this point:

In hostsfile.rb line 17 dscacheutil -flushcache is used to flush the DNS-cache.

In nginx_config.rb the method locate_mime_types finds the file mime.types from the nginx bundle by looking in the folders used by popular OS X package managers (MacPorts, Homebrew)

Also some places I use backticks to call unix commands (like calling kill to signal processes) while I should have used the Process api in ruby. I also use lsof and ps to find running processes and extract data about them. This makes brow vulnerable to the specific formatting of the output from these commands on a particular system. Brow would be more robust if it used more stable api's for this, but I have not researched if that is possible. This is particularly relevant in server_process.rb.

In watcher.rb the gem guard is used to watch the file system. This should be platform independent, but might not be.

Also watcher.rb looks for the command growlnotify to display user notifications. Growl is an OS X specific application, but something similar may exist for linux?¨

Good luck!

Fails to start Origo sometimes [housekeeping branch]

# brow down
Killing nginx
Giving all unicorns a break ...
Stopping hermes, origo, vanilla, checkpoint

# brow up
Releasing master unicorns for checkpoint, hermes, origo, vanilla ...
Launching nginx.
Updating /etc/hosts
Done. Waiting for unicorn workers.
/Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/source.rb:801:in `rescue in load_spec_files': git://github.com/amedia/redcloth.git (at master) is not checked out. Please run `bundle install` (Bundler::GitError)
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/source.rb:799:in `load_spec_files'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/source.rb:381:in `local_specs'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/source.rb:774:in `specs'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/lazy_specification.rb:53:in `__materialize__'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/spec_set.rb:86:in `block in materialize'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/spec_set.rb:83:in `map!'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/spec_set.rb:83:in `materialize'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/definition.rb:113:in `specs'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/definition.rb:158:in `specs_for'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/definition.rb:147:in `requested_specs'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/environment.rb:23:in `requested_specs'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/runtime.rb:11:in `setup'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler.rb:116:in `setup'
    from /Users/alex/.rvm/gems/ruby-1.9.3-p374@global/gems/bundler-1.2.3/lib/bundler/setup.rb:17:in `<top (required)>'
    from /Users/alex/.rvm/rubies/ruby-1.9.3-p374/lib/ruby/site_ruby/1.9.1/rubygems/custom_require.rb:36:in `require'
    from /Users/alex/.rvm/rubies/ruby-1.9.3-p374/lib/ruby/site_ruby/1.9.1/rubygems/custom_require.rb:36:in `require'
  + hermes
  + vanilla
  + checkpoint

Got tired of waiting for origo.
You may try again soon or check syslog for errors.

More automagick

What about a "brow update" that

pulls the latest git repositories in the pebbles
updates/install the bundles
and run the needed rake tasks if needed?

Force use of bash

It seems ruby is hardcoded to use 'sh'. In order to get shell_environment.rb to work as intended, try this:

bash -lc "command"

which should load .bash_profile and execute the command.

Nginx config does not work on Homebrew [housekeeping branch]

Homebrew's Nginx is not compiled with --with-http_gzip_static_module or --with-http_stub_status_module.

nginx: [emerg] unknown directive "gzip_static" in /tmp/brow/nginx/nginx.conf:29
nginx: configuration file /tmp/brow/nginx/nginx.conf test failed

and

nginx: [emerg] unknown directive "stub_status" in /tmp/brow/nginx/nginx.conf:47
nginx: configuration file /tmp/brow/nginx/nginx.conf test failed

Automatic reloading does not work

When running

 $ brow watch

Brow should watch the files of the running servers and reload them when files change. After the latest changes this stopped working. Someone should take a look and make it work again.

"brow down" does not wait for completion [housekeeping branch]

After brow down completes, the Unicorn workers are still working.

Regression: Default app does not work [housekeeping branch]

The old behaviour where a single app handled all non-matching requests is broken.

# echo origo >~/.brow/default
# brow down; brow up
# curl "http://indrepuls.no.dev"
<html><body><h1>503 Service Unavailable</h1>
No server is available to handle this request.
</body></html>

A default backend should have been added to HAProxy config.

Brow stopped watching repos

After the port to Ubuntu brow watch is no longer automatically restarting services. It is not clear to me why this happens. Fedir?

Support capitalized project names

Mappings in the nginx-config is naïvely based on the project folder name in the .brow-folder. The names should probably be downcased before it is passed to nginx.

Cleaner rewriting of /etc/hosts

At some point, implement hosts configuration with this?

https://github.com/bjeanes/ghost

Avoid alias expansion for shell commands

brow does not work if the user has an alias for one of the shell commands it uses, e.g. ps. Use full path (/bin/ps) to avoid alias expansion.

En annen måte er å kjøre "shopt -u expand_aliases" før alle kommandoer. Da vil ikke bash bruke aliaser.

Brow sets (command-line feature)

It would be nice to have some convenience tools to work with brow, for example to create a brow set, i.e. a ~/.project.brow directory unibrow? for a single app with any dependencies (other apps/pebbles).

e.g.
brow new unibrow lifeloop --path ~/code/apps/lifeloop --pebblespath ~/code/pebbles --pebbles checkpoint papertime tiramisu

which would create a ~/.lifeloop.brow dir with symlinks to lifeloop, papertime, tiramisu, and checkpoint

We'd also need brow use unibrow lifeloop which would unlink an existing .brow dir, and ln -s .lifeloop.brow .brow

"brow list" spews "ps" errors [housekeeping branch]

# brow list
  checkpoint (down)
  hermes (down)
  origo (down)
  vanilla (down)
sh: /tmp/brow/haproxy.pid: No such file or directory
ps: option requires an argument -- p
usage: ps [-AaCcEefhjlMmrSTvwXx] [-O fmt | -o fmt] [-G gid[,gid...]]
          [-g grp[,grp...]] [-u [uid,uid...]]
          [-p pid[,pid...]] [-t tty[,tty...]] [-U user[,user...]]
       ps [-L]
  nginx/haproxy (down)

Minor thing: Brow recommends installing a linux-only notification system on Darwin

A small bug makes brow prompt me to install a notification tool other than Growl. Please just recommend one notification system, and only one that works on the machine in question ;-)

Updating all pebbles at once

There would be more convenient to have some cli command which would update all pebbles within .brow directory.
So it could do:

git pull
bundle install
bundle update
bundle exec rake db:nuke RACK_ENV=development

Print self test and documentation endpoint for each pebble

Brow up should sign off with a list of URLs taking you to a page that actually confirms that the pebble is running and also perhaps has its readme (or a link to such).

Empty hosts file.

I founded brow failes when startin with empty hosts file. With this backtrace:

brow up
Releasing all unicorns ...
Nginx allready running, reloading config.
Updating /etc/hosts
/home/robert/.rvm/gems/ruby-1.9.2-p290@brow/gems/brow-0.0.1/lib/brow/hostsfile.rb:11:in `update': undefined method `+' for nil:NilClass (NoMethodError)
    from /home/robert/.rvm/gems/ruby-1.9.2-p290@brow/gems/brow-0.0.1/lib/brow/wrangler.rb:35:in `up'
    from /home/robert/.rvm/gems/ruby-1.9.2-p290@brow/gems/brow-0.0.1/bin/brow:16:in `up'
    from /home/robert/.rvm/gems/ruby-1.9.2-p290@brow/gems/thor-0.14.6/lib/thor/task.rb:22:in `run'
    from /home/robert/.rvm/gems/ruby-1.9.2-p290@brow/gems/thor-0.14.6/lib/thor/invocation.rb:118:in `invoke_task'
    from /home/robert/.rvm/gems/ruby-1.9.2-p290@brow/gems/thor-0.14.6/lib/thor.rb:263:in `dispatch'
    from /home/robert/.rvm/gems/ruby-1.9.2-p290@brow/gems/thor-0.14.6/lib/thor/base.rb:389:in `start'
    from /home/robert/.rvm/gems/ruby-1.9.2-p290@brow/gems/brow-0.0.1/bin/brow:89:in `<top (required)>'
    from /home/robert/.rvm/gems/ruby-1.9.2-p290@brow/bin/brow:19:in `load'
    from /home/robert/.rvm/gems/ruby-1.9.2-p290@brow/bin/brow:19:in `<main>'

Create log directory if it doesn't exist

It would be nice if brow could automatically create the apps' ./log directory if it doesn't exists.

start current project

It would be veeeery nice if we could say

$ brow start .

in order to start the project (and all the dependencies) of the project in the pwd.

Should create log folder unless one exists

Won't launch Nginx because config did not validate. Nginx had this to say about that:
nginx: the configuration file /tmp/brow-nginx-preflight.conf syntax is ok
nginx: [emerg] open() "/home/tom/.brow/.tiramisu.brow/tootsie/log/nginx-access.log" failed (2: No such file or directory)
nginx: configuration file /tmp/brow-nginx-preflight.conf test failed

Warn about pow remnants if such exist

Tell the user to remove the file /etc/resolver/dev (lovingly generated by pow).

Brow should get its app name from the symlink

Brow gets the name for the app from the symlink, but the name for the process from the pwd. Should only get the name from one or the other.