Activity (RSS)

fixed on git head

When starting a cluster like this:

thin start -e production -s 2 -S /var/run/thin.sock --pid /var/run/thin.pid -l /var/log/thin.log -c /u/apps/blog -D 

The ...

That was a fix for Ruby 1.9, I added a condition to still use each under Ruby 1.9.

Fixed on git.

thx for reporting this Nathan

url is now configurable and I also made the page look more pretty.

I still want to add more info on the page.

it looks like it's letting the connections finish their work before returning and stopping the server, but doesn't seem like the data is sent back the the clie...

We should also add the pid and/or port/socket that the thin process is running on.

We should be able to specify a stats uri either by passing it to --stats in the command line via

thin start --stats=/stats

or by including it in the config f...

Are you having this error running a Rails application?

Using thin-0.6.1-x86-mswin32-60 gem:

lib/thin/response.rb - lines 37-41

    def headers=(key_value_pairs)
      key_value_pairs.each do |k, vs|
        vs....

recent benchmarks shows Thin uses less mem then mongrel: http://macournoyer.files.wordpre...

I tried reproducing in any way I could, the error is always caught by Thin's rescue block.

Plz reopen if you can reproduce

some people use monit for ruby web app process management - give examples of usage

I've made the tweak - hard to know if it worked until actually trying to push the next release

Windows doesn't do UNIX sockets, so we should throw an error and exit.

closing as it has been released. UI enhancements can be a different ticket

EM supports a huge number of open connection so DoS is not a problem. Started implementation on keepalive branch will merge and put in next release.

Doesn't seem to stop when processing multiple requests.

We could send a HUP signal that would launch a new thin process w/ the same options.

When restarting thin, it would be nice if it either remembered your startup options or just sent a HUP to the running processes.

Nginx (and maybe other load balancer) support connecting to backend through unix domain socket. Which is the fastest interprocess communication method. EventMa...

Make sure we prevent DoS attacks

http://rubyforge.org/pipermail/m...

http://www.io.com/~maus/HttpKeep...

Proving some load balancing builtin the thin script or very easy to use.

Check possibility to integrate w/ swiftiply or build custom.

Has to be configurable ...

I'll try that thx

Here's the pertinent part of the stack trace:

/usr/local/lib/ruby/gems/1.8/gems/thin-0.5.3/lib/rack/adapter/rails.rb:52:in `serve_rails'

/usr/local/lib/ruby/...

Just added something very basic:

== Server stats

3 requests
3 requests finished
0 errors
25.140546 uptime
Running on thin 0.5.5 codename Pony

== Las...

I can reproduce the error, but in my case, it's caught by the rescue block and the server keeps running.

weird thing is the error is not caught by the rescue Object => e in Connection#process

how come?

--trace now logs request/response and show full backtrace on each error.

Applied modified version of your patch.

thx Matt

Glad to help!

Looking good Matt, I'll take deeper look when I get a sec and apply.

thx a lot!