Ruby and Rails

Some Ruby related executables such as rake seem to be missing and gem gives me this message:

/usr/local/lib/site_ruby/1.8/rubygems.rb:1:in `require': no such file to load -- rbconfig (LoadError)
from /usr/local/lib/site_ruby/1.8/rubygems.rb:1
from /usr/bin/gem:3:in `require'
from /usr/bin/gem:3

I used rake a few nights ago with no problems and also used gem about a week ago.

We're seeing exactly the same thing...not sure why; we've already filed a support ticket for this (literally minutes ago).

I'm having problems with a rails hosted site this morning -

entering rails --version at the command line seems to return command not found, and there are no fcgi instances being spawned. Anyone else seeing this problem?

Same here.

Yep, me too. I see the exact same issues. Are you all on crackerjack?

Yup, rails has stopped working for me. I tried to run rake to unfreeze rails, and rake wasn't found. A quick investigation showed that while 'locate rake' listed rake as being at

/usr/lib/ruby/gems/1.8/bin/rake
it was actually at
/usr/lib/ruby/ruby/gems/1.8/bin/rake

what's that extra ruby doing in the path? I guess the symbolic links for ruby at /usr/bin/ are all dead now, so that's why irb/rake aren't working :(

further edit: i'm on scotch, btw.Edited by mutagene on 05/19/06 06:17 AM (server time).

> Are you all on crackerjack?

Nope, I'm on holt and seeing the same thing.

On gower:

./../config/boot.rb:7:in `require': no such file to load -- pathname (LoadError)
from ./../config/boot.rb:7
from ./../config/environment.rb:11:in `require'
from ./../config/environment.rb:11
from dispatch.cgi:3:in `require'
from dispatch.cgi:3

[gower]$ rails --version
-bash: rails: command not found
[gower]$ ruby -v
ruby 1.8.2 (2005-04-11) [i386-linux]

Same on Hillhurst

Yep, exactly the same problem here on Rollo. Looks as though Ruby is stuffed.

Frustrating considering our site has been down for over 2 hours and no response from Support. I guess a lot of other people must be experiencing the same issues.

yes, same on melrose : my 2 rails powered sites are totally dead and rails --version --> command not found...

Same here, on boyd. Rails and Rake are both missing, while "ruby script/console" returns a 'require' error and my dispatch.fcgi's are timing out.

same on admiralty. hopefully this'll get fixed soon.

even my site is down, it happened 8-9 hour ago, after showing my site for a presentation . suddenly site response failed with rails error.

as usual i shelled to server and killed my dispatch.fcgi, but they did no start back.

could not run console, or deploy from remote(*as it requires some components from my remote site)

Have posted a sitewide error, which got this reply,

Outage resolved: No server-wide problem was found.
(2006-05-19 02:53:37 PDT, 0 secs ago)
Outage first reported.
(2006-05-19 02:31:21 PDT, 22 mins 16 secs ago)



have tested with freezing my rails.

even tried modifications for config/boot.rb ,

have checked ruby folder, gem folders, also error location

no luck yet.

hope the outage is fixed soon, dont want to waste my weekend watching movies

regards
Senthilnayagam

http://senthilnayagam.com/

Someone probably deployed across machines and got a path wrong in a script, thus breaking everything with the duplicate /ruby/ruby

I do find the lack of feedback from Dreamhost on this (and similar issues) a bit frustrating. Isn't this supposed to be what dreamhoststatus.com is for?

Coming after the recent downtime due to the unannounced switch to Rails 1.1, this is going to be difficult to explain to clients. Can't say I'd recommend Dreamhost as a place to host rails sites given the experience of the last few months.

We first noticed the problem a bit after 9PM Thursday on alondra.

Symptoms are as follows:

I can't run 'ruby script/console' without getting:

./script/../config/boot.rb:6:in `require': no such file to load --
pathname (LoadError)
from ./script/../config/boot.rb:6
from script/console:2:in `require'
from script/console:2

Even when I go find the 'rake' that's on the weird path at
'/usr/lib/ruby/ruby/gems/1.8/bin/rake' it blows up with:

/usr/local/lib/site_ruby/1.8/rubygems.rb:1:in `require': no such file
to load -- rbconfig (LoadError)
from /usr/local/lib/site_ruby/1.8/rubygems.rb:1
from /usr/lib/ruby/ruby/gems/1.8/bin/rake:9:in `require'
from /usr/lib/ruby/ruby/gems/1.8/bin/rake:9

And finally my Rails web application does run. When I try to manually
run its dispatch.fcgi in ~/lentigen_web_site/public/dispatch.fcgi I
get (unsurprisingly in light of the above)...

./../config/boot.rb:6:in `require': no such file to load -- pathname
(LoadError)
from ./../config/boot.rb:6
from ./../config/environment.rb:8:in `require'
from ./../config/environment.rb:8
from ./dispatch.fcgi:21:in `require'
from ./dispatch.fcgi:21

The weird double '.../ruby/ruby/...' in that one path, plus a broken symlink make me think that perhaps someone ran a script from somewhere they weren't supposed to...

I hope this is fixed soon. We were supposed to demo a system to a customer this morning on our staging machine, and they just signed up for a Strictly Business account at our recommendation a few days ago. We're about a week from wanting to go live on that machine, but at the moment nothing works and it's been 12 hours with no indication that anybody's aware of or working on the problem... :-(

Hello folks,

Here's a workaround...

export PATH="/usr/local/lib/ruby/ruby/gems/1.8/bin:$PATH"
export RUBYLIB="/usr/local/lib/ruby/ruby/1.8:/usr/local/lib/ruby/ruby/1.8/i386-linux"
export GEM_HOME="/usr/local/lib/ruby/ruby/gems/1.8"

For Rails, you can prepend the following to your config/boot.rb:

stuff = %w(/usr/local/lib/ruby/ruby/1.8 /usr/local/lib/ruby/ruby/1.8/i386-linux)

$LOAD_PATH.unshift stuff[0]
$LOAD_PATH.unshift stuff[1]

ENV["RUBYLIB"] = stuff.join(":") # Needed for script/console
ENV["GEM_HOME"] = "/usr/local/lib/ruby/ruby/gems/1.8"

Thanks for the effort! What really needs to happen though is a re-deploy that fixes the last one's path screwup.

Thanks. Yeah, I wonder why it's taking so long. Being a relatively new DreamHost customer this breakage (and some other issues) make me a bit angry.

And of course some communication and a warning that the fix is being applied so that anybody who uses Qerub's path magic doesn't then immediately break when the weird Little Ceasar's Commercial-like "/ruby/ruby" paths go back to normal. :-)

I tried implementing this fix, and it seems to partially solve the problem. I can run dispatch.fcgi locally without errrors, and running script/console works again. However, my site is still not responding. Any idea what could be missing? My ruby installation is frozen, do I need to change the gem path to include my frozen rails directory?

Same issue on edendale. I had a demo scheduled today, which, as you can imagine, didn't go well with a missing site.

RubyGems doesn't support having more than one gem home at the same time, so I can't provide a workaround that won't have any effect once Ruby is restored to a working state. Sure, one could wrap require/require_gem but that's the kind of magic I won't touch. This said, if your Rails application doesn't depend on RubyGems, you can apply my workaround without breaking the app in the normal state.

Oh, right. I had the same issue and ended up switching to CGI mode instead of FastCGI. I might take a look at it after dinner, since CGI is pretty sub-optimal...

ahh, ok, so now I need to back out these environment changes, what should the values of the RUBYLIB and GEM_HOME variables be (I know I should've checked before I changed them, whoops!)

It is now working again on LARCHMONT and KARNOV.

They fixed the /usr/lib/ruby/ruby to just be /usr/lib/ruby

Everything now works.

FINALLY.

Working on alondra too.

Cienega as well, just for the record.

On Alvarado, is anyone noticing a SLOW (i.e. greater than 4-5 minutes) response time for ruby apps? It's not completely fixed for me.

I are really sorry about the ruby downtime, it was a preventable error that should have been caught in testing. I also apologize that the problem was not able to be solved by the night staff and continued as long as it did.

I found alvarado to be suffering from a users runaway script that was causing a high load on the server. Please send in a support request if you experience any slow downs on a particular server. If you post about an issue here, it is good to include your domain as well so that I can locate your support request. thanks.

I have problems with Rails, too. I just got my account at Dreamhost and installed typo on there yesterday and all was okay, besides that the server is really slow I just get about 30 kb per second (I tried downloading a file yesterday). I followed the instructions in the dreamhost wiki about typo and the instructions on the typosphere trac site about typo and dreamhost. As of today, the site no longer works. It loads and loads and after a few minutes it has finished loading, but only the typo frontpage. If I try to get into the admin section, after a few minutes of loading I get a "Typo failed to start properly" error message. I then tried to incorporate the additional fixes for typo on dreamhost, those for fixing the error 500 messages and did a killall -9 on dispatcher.fcgi, but it didn't help with the problem. Any suggestions? Server is Pez, website is blog.fanfictionality.com