Activity (RSS)

Well, here's the patch. I've also added reload support to the runner, so you can do:

thin reload --servers 3 -e production

and it'll go through all of them g...

Mongrel does it w/ the HUP signal but it's not working very well.

As it's written when Mongrel boots:

Rails signals registered.  HUP => reload (without resta...

I'm not sure if this is a good idea (or possible) and I'm not sure if this is the right forum for it.

However, wouldn't it be nice to deploy without having to...

I'll try to prototype that this weekend.

Want concerns me is the extra memory required to run a master process. An empty Ruby VM can use 80 MB on some system....

duplicate of http://thin.lighthouseapp.com/pr...

>> Writing PID to tmp/pids/thin.0.pid
>> Thin web server (v0.6.3 codename Ninja Cookie)
>> Listening on /tmp/thin.0.sock, CTRL+C to stop
Unexpected error whi...

Fixed on http://github.com/macournoyer/th...

thin -s20 -eproduction --socket /tmp/thin.sock -V stop

thin.0.log:

/usr/local/lib/ruby/gems/1.8/gems/eventmachine-0.11.0/lib/eventmachine.rb:512:in `stop_tcp...

Should be fixed in http://github.com/macournoyer/th...

You can install the beta gem:

sudo gem install thin --source http://code.macournoyer.com

Let me know...

thin -s20 -eproduction --socket /tmp/thin.sock -V start

thin.0.log:

>> Writing PID to tmp/pids/thin.0.pid

>> Thin web server (v0.6.3 codename Ninja Cookie)

...

It would be great to be able to run thin under a service manager such a daemontools/runit/launchd/AIX's SRC. The advantages of such an approach are:

• Unified ...

I run thin through config file:

pid: /var/run/thin_cluster/myapp.pid
log: log/thin.log
timeout: 60
port: 4660
chdir: /var/www/apps/myapp/current
environm...

Can you tell me exactly what commands you ran

Sometimes when I restarted my cluster I have this error that I can't fix.

>> Writing PID to /var/www/apps/italyabroad/current/tmp/pids/thin.4360.pid
>> Thin ...

http://groups.google.com/group/e...

3 thin servers crashed and the following error was in the log files:

terminate called after throwing an instance of 'std::runtime_error'
  what():  unable to...

Fixed in http://github.com/macournoyer/th...

Well, the spawned clusters talk to the Runner via a socket and tell it if it was successful or not, passing along error information if not... so the Runner can...

You mean pinging the host until is response? Not sure I understand your solution.

Or maybe having the child process output the log on error.

Not necessarily the best idea, but if you had the spawner open up x number of ports (x being the number of children to spawn) where each child connects to the ...

(Sorry for the double post... apparently had network problems when I posted it so didn't realize it went through the first time.)

Seems good to me, considering it's a bit more forward-thinking so if you have other options you need to exclude later on, it's easy to update.

Nice and simple...

Fixed in http://github.com/macournoyer/th...

I removed the code deleting the options and just removed them before sending it as a command to individual server...

The problem is that Controllers::Cluster#initialize deletes the :servers key from @options and stores it in @size.

One potential solution would be to not dele...

The problem is that Controllers::Cluster#initialize deletes the :servers key from @options and stores it in @size.

One potential solution would be to not dele...

/me applauds excitedly!

applied w/ some modifications, now on git master

thx a lot for the contribution Alex!

nice catch James, seems to be the only option not saved to the config file. I'll fix that.

I'm running thin 0.6.3 on Ubuntu.

When I run `thin config --server 3 -C /etc/thin/myapp.yml`, I get:

---

pid: tmp/pids/thin.pid

log: log/thin.log

timeout:...

This is 'cause -D set $DEBUG=true which make Ruby more verbose and less tolerable.

I'll review this option to be more useful and maybe merge w/ -V option.

th...

On a blank rails 2.0.2 app, I create a single controller and an index view for it.

In this view I put :

<%= javascript_include_tag 'application' %>

and whe...

fixed on git

thin restart --servers 1

-s1 should be considered as a cluster.

Reported by cnantais on irc channel

This is the same behaviour as mongrel_cluster so I'm moving to 'Future'.

It would require waiting for a couple sec and checking it the process is still runnin...

it's on git now:

Add the --rackup option to load a Rack config file instead of the Rails adapter.

So you can use any framework with the thin script and start...

Duplicate of http://thin.lighthouseapp.com/pr...

fixed on git

thx for the suggestion Daddye, feels more intuitive now!

I updated the example/thin.god file w/ this one on git.

thx Gump!

You're right, I'll include it in the example dir for now.

Wow, this is freaking awesome Gump!

I'd really love to integrate this into thin. What's your idea about this, do you think we should replace runlevel service ...

to be tested and benchmarked

A recent fix to Ruby 1.9 GC might save 1/3 of mem usage when using fork to start thin servers.

http://izumi.plan99.net/blog/ind...

Example script to prefork ...

Is not a problem at the moment manually do this change but if not necessary check the directory when thin make the config you help my hands

hehe

Super thanks...

-c change the current directory so that all other paths can be relative to it.

Not sure this make sense when generating the config file.

But you could:

thin...

When I do:

thin config -c /my/app/path/*current* -C /my/config/dir/app.conf

thin check if /my/app/path/current exist... why?

I ask to remove the check of a...