Capistrano Google Group

Re: [capistrano] Re: Capistrano seemingly freezes on deploy (used to work fine)

robin-li...@robinbowes.com (Robin Bowes) — Sat, 12 Dec 2009 11:16:27 UT

Indeed, I would argue that this is a better approach and should be made
the default.
For example, it means that you don't have to have subversion installed
on your production servers, and only your deployment server needs access
to your subversion repository.
In fact, I don't personally see *any* benefit in deploying with :svn

Re: [capistrano] Re: Capistrano seemingly freezes on deploy (used to work fine)

r...@deadcrow.net (Rory Gibson) — Sat, 12 Dec 2009 10:30:47 UT

If you've got SVN connection issues on the remote server, don't forget
that you can work around them by using the :copy strategy, which will
do a local checkout on the machine where ca is running, create a
tarball, ship it over and unack into the familiar releases/ & current/
directory structure.
See [link] for more info.

Re: Capistrano seemingly freezes on deploy (used to work fine)

webch...@gmail.com (antum) — Sat, 12 Dec 2009 09:57:08 UT

Hi Rafael,
Thanks for your help, I checked out the docs for the setting, and
tried it out but to no avail, see my reply to Lee for more
information.
Cheers
Ant

Re: Capistrano seemingly freezes on deploy (used to work fine)

webch...@gmail.com (antum) — Sat, 12 Dec 2009 09:55:26 UT

Hi Lee,
Many thanks for your reply, I've only just had a chance to sit down in
front of the computer. So...
Is remote_cache enabled by default? I don't have it set anywhere in my
deploy.rb
I just tried a checkout locally, it took 5 minutes, I tried the same
thing from the server and it's still going after 32 minutes...

Re: [capistrano] Capistrano seemingly freezes on deploy (used to work fine)

r...@aspgems.com (Rafael G.) — Fri, 12 Dec 2009 15:59:42 UT

Hi Ant,
Could you try to add this line to your deploy.rb ?
default_run_options[:pty] = true
More info about this change:
:pty option - [link]

Regards

Re: [capistrano] Capistrano seemingly freezes on deploy (used to work fine)

lee.hamb...@gmail.com (Lee Hambley) — Fri, 12 Dec 2009 09:37:28 UT

Ant, first thing that comes to mind is a repository problem; whenever people
have problems – we usually recommend that they don't use :remote_cache for
deploying, but it might work for you.
My suggestion, before you try that would be to manually go onto the server;
crank up the log level and try a checkout from your repository. Also, check

Capistrano seemingly freezes on deploy (used to work fine)

webch...@gmail.com (antum) — Fri, 12 Dec 2009 01:56:47 UT

Hi there,
I'm trying to deploy a PHP application from SubVersion source control
to a single server. The recipe hasn't changed and it used to work
fine, but now it stops after doing a svn checkout and doesn't budge
for hours until I press Cmd+C (I'm in Terminal, on a Mac)
See [link] for my Terminal output. (You can see

Re: Newcomer Problem

mafor...@gmail.com (mafordha) — Thu, 12 Dec 2009 00:34:44 UT

'deploy_via, :copy' did the trick, thanks so much!!!

Re: [capistrano] Re: Deploying Error: `split' called for 440:Fixnum

lee.hamb...@gmail.com (Lee Hambley) — Wed, 12 Dec 2009 16:04:39 UT

Memo, thanks - I spoke with a friend of yours on IRC – I will try to package
a debug version of Cap which can give you better output.

-- Lee Hambley

Twitter: @leehambley | @capistranorb
Blog: [link]
Working with Rails: [link]

2009/12/16 Memo <m...@52nineteen.com>

Re: Deploying Error: `split' called for 440:Fixnum

m...@52nineteen.com (Memo) — Wed, 12 Dec 2009 15:59:49 UT

I tried using the same test using my mac book and came up with the
same error.

[link]

Re: Deploying Error: `split' called for 440:Fixnum

m...@52nineteen.com (Memo) — Wed, 12 Dec 2009 15:26:13 UT

It did not work. So is the problem on my machine. HostingRails said
they have nothing block but I'm starting to wonder.

Results: [link]

Magnus is hosted and supported by HostingRails. From their response,
they believe that the cap commands are run from my local machine into

Re: [capistrano] Re: Deploying Error: `split' called for 440:Fixnum

lee.hamb...@gmail.com (Lee Hambley) — Wed, 12 Dec 2009 15:13:28 UT

Can you get a shell on your server even with this? Can you explain
"HostingRails says that it is a problem when unfuddle tries
to connect to their magnus server." - Unfuddle is also managing your
`magnus` server?

Try this:

$ cd ~
$ mkdir cap_test
$ cd cap_test
$ echo "role :web, 'my-server.com'" > Capfile

Re: Deploying Error: `split' called for 440:Fixnum

m...@52nineteen.com (Memo) — Wed, 12 Dec 2009 15:05:30 UT

I tried it. I ran a "cap deploy:setup" and "cap deploy:check" but
still came up with the same errors. If I do a cap deploy:setup or cap
deploy:restart where it doesn't need unfuddle, is it executing from my
machine. HostingRails says that it is a problem when unfuddle tries
to connect to their magnus server.

Re: [capistrano] Re: Rollback deletes wrong release

lee.hamb...@gmail.com (Lee Hambley) — Wed, 12 Dec 2009 15:00:39 UT

The solution is to figure out the way that works for your infrastructure to
`ls` them in the right way, then you can overload this default:

set(:releases) { capture("ls -xt #{releases_path}").split.rever se }

Defined in:
[link]

Re: [capistrano] Re: Deploying Error: `split' called for 440:Fixnum

lee.hamb...@gmail.com (Lee Hambley) — Wed, 12 Dec 2009 14:54:05 UT

I suspect now then that this is a connection problem from your server, to
reach your repository host; please try one thing for me

set :deploy_via, :copy

Then, no such connection is required (hardly a fix, I know - but we can
debug some more!)

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Re: [capistrano] Newcomer Problem

lee.hamb...@gmail.com (Lee Hambley) — Wed, 12 Dec 2009 13:30:30 UT

If you have no remote repository, you need to set deploy_via, :copy - and
jump through some more hoops.

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Newcomer Problem

mafor...@gmail.com (mafordha) — Wed, 12 Dec 2009 06:35:48 UT

Hello all,

Ruby 1.8.6, Capistrano v2.5.10, Git version 1.6.5.1

I am attempting to get Capistrano setup and working for the first
time.
This is very new to me. I am trying to deploy from a local Git
repository to my shared host.

After running 'cap deploy:setup' and 'cap deploy:check' everything

Re: [capistrano] Re: Best Reaper Approach ??

lee.hamb...@gmail.com (Lee Hambley) — Wed, 12 Dec 2009 08:37:09 UT

That depends on about a million factors. Apache is easier (read: better
documented) - don't optimise by experimenting with nginx until you see what
your performance problems are.

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Re: Rollback deletes wrong release

streetda...@gmail.com (booyah) — Wed, 12 Dec 2009 07:28:02 UT

nb. I've found that run'ing any command that modifies any release
folder or its contents will cause the dodgy timestamp to appear
again. So be careful what you do in a :before_update_code task...

Re: Deploying Error: `split' called for 440:Fixnum

m...@52nineteen.com (Memo) — Wed, 12 Dec 2009 01:48:55 UT

I updated the deploy.rb file and ran both cap deploy:setup and cap
deploy:check. They both had the same connection issue.

[link]

Re: Rollback deletes wrong release

streetda...@gmail.com (booyah) — Tue, 12 Dec 2009 23:31:29 UT

I think I've been able to solve it by touched the latest release
folder after update_code:

task :after_update_code, :roles => [:web] do
# Sleep may not be needed
run "sleep 1"
run "touch #{latest_release}"
end

I deleted all releases and did a number of deploys and rollbacks and

Re: Rollback deletes wrong release

streetda...@gmail.com (booyah) — Tue, 12 Dec 2009 22:58:02 UT

I've come to the same conclusion. I'm not using file-based rails
caching. I'm just using it for deploying a PHP site and have
overridden a lot of the rails specific tasks in deploy.

The ordering of the releases from 'ls -xt' is the problem, but the
reason I think they're in the wrong order is because of the

Re: Best Reaper Approach ??

akl...@gmail.com (Arco) — Tue, 12 Dec 2009 22:57:54 UT

OK I'll give passenger a try.

From the passenger docco, i see that you can run passenger with either
apache or nginx.

Which one is better ??

Re: [capistrano] Best Reaper Approach ??

lee.hamb...@gmail.com (Lee Hambley) — Tue, 12 Dec 2009 22:40:47 UT

Honesty deploy on Passenger or something else now, Mongrel is out! (We're
actually removing support in the new year, it'll be available as a plugin
after that)

If you are goig to use the reapers (I have no experience) - but I suspect
you need that plugin (otherwise, the scripts don't exist)

Best Reaper Approach ??

akl...@gmail.com (Arco) — Tue, 12 Dec 2009 22:28:13 UT

i have a question re: using the reaper script

i'm deploying a new app for the 1st time, using the newest version of
rails and capistrano

should i use the 'old style' reaper (I saw a plugin for this)

or is there a better replacement for the old style reaper ??

Re: [capistrano] Re: Deploying Error: `split' called for 440:Fixnum

lee.hamb...@gmail.com (Lee Hambley) — Tue, 12 Dec 2009 21:53:06 UT

Disable cached copy, and switch to :export or :checkout. Rinse and repeat,
did you post this problem previously under another name, its really
familiar?

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Re: Deploying Error: `split' called for 440:Fixnum

m...@52nineteen.com (Memo) — Tue, 12 Dec 2009 21:48:22 UT

It is the entire backtrace. Here is the entire backtrace for cap
deploy.

[link]

The software has been pushed to the remote repository. In fact, I
have multiple changes that have been committed and each shows up in
unfuddle and github when I pushed it there.

Thanks for the help.

Re: [capistrano] Re: Deploying Error: `split' called for 440:Fixnum

lee.hamb...@gmail.com (Lee Hambley) — Tue, 12 Dec 2009 21:42:19 UT

And the whole backtrace please?

Also are you certain that the git branch you are deploying has been pushed
to your remote repository ?

( no need to try and embed the things, just paste the links )

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Re: Deploying Error: `split' called for 440:Fixnum

m...@52nineteen.com (Memo) — Tue, 12 Dec 2009 21:39:48 UT

I haven't really used this so hope it works.

Capfile
<script src='[link]></script>

deploy.rb
<script src='[link]></script>

Shell
<script src='[link]></script>

Re: [capistrano] checking the dry-run flag from within a task

lee.hamb...@gmail.com (Lee Hambley) — Tue, 12 Dec 2009 21:32:56 UT

I'm not sure it ever came up that anyone asked about the "local" thing for
use with dry-run, I honestly have never used it, I always make my calls with
pure ruby. Even then I'm not sure whether or not it honors it.

The FileUtils class should provide the behavior you need, maybe in this way
you can have cap honor the dry run flag, either way - let us know how you

Re: [capistrano] Deploying Error: `split' called for 440:Fixnum

lee.hamb...@gmail.com (Lee Hambley) — Tue, 12 Dec 2009 21:27:19 UT

Please paste code in a Gist/Pastie, and your backtrace (in its entirety)

-- Lee Hambley

Twitter: @leehambley | @capistranorb

checking the dry-run flag from within a task

pharri...@gmail.com (Pharrisee) — Tue, 12 Dec 2009 19:27:46 UT

Lo,

I've got a highly customised set of tasks that have both local and
remote parts to them, when I use the dry-run flag the remote tasks
honour the flag, but, obviously, the local tasks don't honour it. I
understand why it is but I would like to do something to change this.

I've created a normal ruby method which runs all my local stuff:

Deploying Error: `split' called for 440:Fixnum

m...@52nineteen.com (Memo) — Tue, 12 Dec 2009 16:28:30 UT

This error was posted before but the solution of going to capistrano
2.5.5 did not work for me.

I am getting this error and I've tried several different recipes. It
appears to be a problem connecting to my host server. Right now I am
running on OS X10.6.1 with capistrano 2.5.10 and net-ssh 2.0.17. I

Re: [capistrano] Rollback deletes wrong release

lee.hamb...@gmail.com (Lee Hambley) — Tue, 12 Dec 2009 09:56:30 UT

Are you using file-based rails caching? We have an open bug for that, in a
nutshell the `ls -xt` command we use, lists them in the `wrong` order if you
are using file caching.

( this could be an easy fix, but how should we approach it? – thus we
haven't )

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Rollback deletes wrong release

streetda...@gmail.com (booyah) — Tue, 12 Dec 2009 01:24:32 UT

I've been introduced to Capistrano by a colleague and must say i'm
loving it, however, I'm seeing some strange behaviour with the
rollback task:

for example, lets say I have 3 releases with 1 being the newest and 3
being the oldest. If I do cap deploy:rollback I would expect it to
roll back from release 1 to release 2 and release 1 would be deleted.

Re: Deploy from git repository with multiple projects

sc...@scottjohnson.org (Scott Johnson) — Fri, 12 Dec 2009 18:24:49 UT

On Dec 10, 9:47 am, theturingmachine <theturingmach...@gmail.com>
wrote:

I think there's a more fundamental issue here. This might be more work
than you want to take on, but you should really think about
restructuring these apps into independent repos. Shared code should go
into plugins or gems, each with their own repo.

Re: [capistrano] Deploy from git repository with multiple projects

lee.hamb...@gmail.com (Lee Hambley) — Thu, 12 Dec 2009 20:15:30 UT

Rory, great - I hope there's a contribution in it somewhere! It would be a
real shame not to share community knowledge – there's not enough people
hacking on the internals like you are doing!

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Re: [capistrano] Deploy from git repository with multiple projects

r...@deadcrow.net (Rory Gibson) — Thu, 12 Dec 2009 19:49:18 UT

On 10 Dec 2009, at 17:47, theturingmachine

I've just done something very similar.

All I did was create a capistrano/modules/recipes structure alongside
my deploy.rb and copy the Copy strategy recipe sourcefile into it from
the unpacked gem.

A swift rename later and I had my own strategy, which I then

Re: [capistrano] Deploy from git repository with multiple projects

lee.hamb...@gmail.com (Lee Hambley) — Thu, 12 Dec 2009 19:55:49 UT

I'm afraid this is basically impossible (the Git guys argument is that
repositories are so cheap, you should not be sharing them) - they mean cheap
like overhead, not like hosting charges, but its a fundamental limitation of
Git that you can't easily deploy a "subdirectory" from a repository.

If you are using the :remote_cache deployment strategy, then you can use the

Deploy from git repository with multiple projects

theturingmach...@gmail.com (theturingmachine) — Thu, 12 Dec 2009 17:47:02 UT

i'm currently migrating a project we had on svn to git. the conversion
has been working fine but now we have to change our deploy recipes to
use git, the problem I'm having is with our current setup.

We have only one repostory where we have several apps. what we're
currently doing checking out one of the directories from the

Re: [capistrano] Net::SSH 2.0.16 Net::SSH::AuthenticationFailed

lee.hamb...@gmail.com (Lee Hambley) — Thu, 12 Dec 2009 06:25:15 UT

The Correct Syntax (straight from the RDocs) is :

Net::SSH.start("host", "user", :password => "password") do |ssh|
result = ssh.exec!("ls -l")
puts result
end

(you send the username as a parameter, password should be part of the
options hash.)

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Net::SSH 2.0.16 Net::SSH::AuthenticationFailed

lhughe...@gmail.com (lee) — Thu, 12 Dec 2009 02:38:34 UT

I was told folks on this mailing list might be able to help me with
this problem.

I'm running ruby 1.9.1p243 and net-ssh 2.0.16 and getting this error
trying to connect to a vanilla openssh server on the same LAN:

lee@lee-hp:~$ irb
irb(main):001:0> require 'net/ssh'
=> true
irb(main):002:0> Net::SSH.start

Re: [capistrano] Weird error when running deploy script with gateway

lee.hamb...@gmail.com (Lee Hambley) — Tue, 12 Dec 2009 17:33:29 UT

Please paste all the information we ask for on the group homepage, (use your
imagination, we don't need all of it for this)

Anything else strange about your configuration?

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Weird error when running deploy script with gateway

man...@acceleweb.com (mrmanishs) — Tue, 12 Dec 2009 16:16:35 UT

I am trying to run a deploy script with a gateway, and I'm getting a
weird error. I'm not sure if it's the script or my server
configuration. Does anyone have any hints?

servers: ["qa"]
* establishing connection to gateway 'xx.xx.xx.xx:9022'
* Creating gateway using xx.xx.xx.xx:9022
* establishing connection to `qa' via gateway

Re: [capistrano] Re: Task execution based on :primary flag

lee.hamb...@gmail.com (Lee Hambley) — Tue, 12 Dec 2009 17:27:55 UT

Either/or 0 lets take the dicussion to the lighthouse app though -- I think
something like another option to task would be nice and clean:

task :whatever, :raise_on_empty_roles => false do
# whatever
end

-- Lee Hambley

Re: [capistrano] Re: Task execution based on :primary flag

donno...@gmail.com (Donovan Bray) — Tue, 12 Dec 2009 16:44:44 UT

That's a good point Lee, at the time I was in a rush and I just
scratched my own itch.

I'll review that monkey patch and wrap the functionality with a
variable so that you can enable it with something like:

set :allow_empty_roles, true

in your deploy.rb

Do you prefer pull requests or patches?

Re: Task execution based on :primary flag

justin.di...@gmail.com (jmadtech) — Tue, 12 Dec 2009 01:01:44 UT

Donovan,

You're a life saver! That solved the problem perfectly.

Thanks again!

Re: [capistrano] Re: Task execution based on :primary flag

lee.hamb...@gmail.com (Lee Hambley) — Mon, 12 Dec 2009 21:38:35 UT

Donoval,

You didn't feel like submitting a patch?

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Re: [capistrano] capistrano deployment

lee.hamb...@gmail.com (Lee Hambley) — Mon, 12 Dec 2009 21:37:51 UT

Josh, the deploy branch made during deploy of a remote cache is normal, and
should be switched to master when the deploy is done.

-- Lee Hambley

Twitter: @leehambley | @capistranorb

Re: [capistrano] capistrano deployment

donno...@gmail.com (Donovan Bray) — Mon, 12 Dec 2009 20:36:59 UT

I've had this happen to me, I found that it was the remote_cache had
been corrupted. Deleting the remote_cache on the server and
redeploying resolved the problem for me.