I've added new HDD to the server and need to use it to save all users info, so I need to move public directory to new HDD.
I don't know what is the best approach in my case, but I decided to change the public directory to be symbolic link pointing to new HDD path.
Based on that I created the below capistrano task to run after deploy:
# I'm using capistrano 3.1
namespace :public_to_symbolic do
desc 'Change public directory to symbolic link pointing to other Hdd'
task :change do
on roles(:app) do
is_directory = File.directory?("#{release_path}/public")
is_symbolic_link = File.symlink?("#{release_path}/public")
is_new_directory_exist = File.directory?('/mnt/newhdd/public')
can_change = is_directory && !is_symbolic_link && is_new_directory_exist
puts "is_directory : #{is_directory}" # prints false
puts "is_symbolic_link : #{is_symbolic_link}" # prints false
puts "is_new_directory_exist : #{is_new_directory_exist}" # prints false
puts "can_change: #{can_change}" # prints false
puts "release_path: #{release_path}" # prints /var/www/myapp/releases/20180305112922
if can_change
puts 'Changing public directory to symbolic link'
execute "mv #{release_path}/public #{release_path}/public_linked"
execute "ln -s /mnt/newhdd/public #{release_path}/public"
execute "rm -rf #{release_path}/public_linked"
end
end
end
end
after 'deploy', 'public_to_symbolic:change'
So, as commented, is_directory, is_symbolic_link and is_new_directory_exist are all returns false, but when I'm doing the same check (i.e File.directory?("#{Rails.root}/public") or File.directory?("/var/www/myapp/current/public")) using Rails console on deployment server, I'm getting true and I'm able to see the public directory!
To avoid this issue I tried also to use current_path instead of release_path but still same results.
I have two questions:
Why is_directory and is_new_directory_exist are always false while task is running?
If there is any better approach to use public directory on different Hdd, please advice.
All suggestions are welcome
All Ruby methods that you use in your Capistrano tasks run on your local machine.
For example:
File.directory?
File.symlink?
These are always evaluated on your local filesystem. Capistrano never runs Ruby code on the remote server. These methods always return false for you because they are trying to find e.g. "#{release_path}/public" on your local computer, which of course does not exist.
To run code on the server, the tools available to you are Capistrano's test and execute methods. These take in a command string that is executed remotely via SSH.
If you want to test if a remote path is a directory, you cannot use Ruby; you have to use something that can be run in a remote shell. Here is one way to test if a path is a directory, for example:
is_directory = test("[ -d #{release_path}/public ]")
Likewise, to test if a path is a symlink:
is_symbolic_link = test("[ -h #{release_path}/public ]")
I keep receiving this error message from the console when visiting my website:
font from origin 'https://xxx.cloudfront.net' has been blocked from loading by Cross-Origin Resource Sharing policy: No 'Access-Control-Allow-Origin' header is present on the requested resource. Origin 'https://www.example.com' is therefore not allowed access.
I've tried everything:
I've installed the font_assets gem
configured the application.rb file
config.font_assets.origin = 'http://example.com'
Whitelisted Headers on Cloudfront as explained in this article to
Access-Control-Allow-Origin
Access-Control-Allow-Methods
Access-Control-Allow-Headers
Access-Control-Max-Age
But nothing, zero, nada.
I'm using Rails 4.1 on Heroku.
This was an incredibly difficult issue to deal with, for two reasons:
The fact that CloudFront is mirroring our Rails app’s response headers requires you to twist your mind around. The CORS protocol is hard enough to understand as it is, but now you have to follow it at two levels: between the browser and CloudFront (when our Rails app uses it as a CDN), and between the browser and our Rails app (when some malicious site wants to abuse us).
CORS is really about a dialog between the browser and the 3rd-party resources a web page wants to access. (In our use-case, that’s the CloudFront CDN, serving assets for our app.) But since CloudFront gets its Access-Control response headers from our app, our app needs to serve those headers as if it is CloudFront talking, and simultaneously not grant permissions that would expose itself to the type of abuse that led to the Same-Origin Policy / CORS being developed in the first place. In particular, we should not grant * access to * resources on our site.
I found so much outdated information out there -- an endless line of blog posts and SO threads. CloudFront has improved its CORS support significantly since many of those posts, although it is still not perfect. (CORS should really be handled out-of-the-box.) And the gems themselves have evolved.
My setup: Rails 4.1.15 running on Heroku, with assets served from CloudFront. My app responds to both http and https, on both "www." and the zone apex, without doing any redirection.
I looked briefly at the font_assets gem mentioned in the question, but quickly dropped it in favor of rack-cors, which seemed more on point. I did not want to simply open up all origins and all paths, as that would defeat the point of CORS and the security of the Same-Origin Policy, so I needed to be able to specify the few origins I would allow. Finally, I personally favor configuring Rails via individual config/initializers/*.rb files rather than editing the standard config files (like config.ru or config/application.rb) Putting all that together, here is my solution, which I believe is the best available, as of 2016-04-16:
Gemfile
gem "rack-cors"
The rack-cors gem implements the CORS protocol in a Rack middleware.
In addition to setting Access-Control-Allow-Origin and related headers on approved origins, it adds a Vary: Origin response header, directing CloudFront to cache the responses (including the response headers) for each origin separately. This is crucial when our site is accessible via multiple origins (e.g. via both http and https, and via both "www." and the bare domain)
config/initializers/rack-cors.rb
## Configure Rack CORS Middleware, so that CloudFront can serve our assets.
## See https://github.com/cyu/rack-cors
if defined? Rack::Cors
Rails.configuration.middleware.insert_before 0, Rack::Cors do
allow do
origins %w[
https://example.com
http://example.com
https://www.example.com
http://www.example.com
https://example-staging.herokuapp.com
http://example-staging.herokuapp.com
]
resource '/assets/*'
end
end
end
This tells the browser that it may access resources on our Rails app (and by extension, on CloudFront, since it is mirroring us) only on behalf of our Rails app (and not on behalf of malicious-site.com) and only for /assets/ urls (and not for our controllers). In other words, allow CloudFront to serve assets but don't open the door any more than we have to.
Notes:
I tried inserting this after rack-timeout instead of at the head of the middleware chain.
It worked on dev but was not kicking in on Heroku, despite
having the same middleware (other than Honeybadger).
The origins list could also be done as Regexps.
Be careful to anchor patterns at the end-of-string.
origins [
/\Ahttps?:\/\/(www\.)?example\.com\z/,
/\Ahttps?:\/\/example-staging\.herokuapp\.com\z/
]
but I think it’s easier just to read literal strings.
Configure CloudFront to pass the browser's Origin request header on to our Rails app.
Strangely, it appears that CloudFront forwards the Origin header from the browser to our Rails app regardless whether we add it here, but that CloudFront honors our app’s Vary: Origin caching directive only if Origin is explicitly added to the headers whitelist (as of April 2016).
The request header whitelist is kind of buried.
If the distribution already exists, you can find it at:
https://console.aws.amazon.com/cloudfront/home#distributions
select the distribution
click Distribution Settings
go to the Behaviors tab
select the behavior (there will probably be only one)
Click Edit
Forward Headers: Whitelist
Whitelist Headers: Select Origin and click Add >>
If you have not created the distribution yet, create it at:
https://console.aws.amazon.com/cloudfront/home#distributions
Click Create Distribution
(For the sake of completeness and reproducibility, I'm listing all the settings I changed from the defaults, however the Whitelist settings are the only ones that are relevant to this discussion)
Delivery Method: Web (not RTMP)
Origin Settings
Origin Domain Name: example.com
Origin SSL Protocols: TLSv1.2 ONLY
Origin Protocol Policy: HTTPS only
Default Cache Behavior Settings
Viewer Protocol Policy: Redirect HTTP to HTTPS
Forward Headers: Whitelist
Whitelist Headers: Select Origin and click Add >>
Compress Objects Automatically: Yes
After changing all these things, remember that it can take some time for any old, cached values to expire from CloudFront. You can explicitly invalidate cached assets by going to the CloudFront distribution's Invalidations tab and creating an invalidation for *.
If you run Rails on Passenger and Heroku: (if not, jump straight to Noach Magedman's answer)
Noach Magedman's answer was very useful for me to set up CloudFront properly.
I also installed rack-cors exactly as described and whilst it worked fine in development, the CURL commands in production never returned any of the CORS configurations:
curl -H "Origin: https://tidyme-staging.com.au" -I http://tidyme-staging.com.au/assets/31907B_4_0-588bd4e720d4008295dcfb85ef36b233ee0817d7fe23c76a3a543ebba8e7c85a.ttf
HTTP/1.1 200 OK
Connection: keep-alive
Server: nginx/1.10.0
Date: Wed, 03 Aug 2016 00:29:37 GMT
Content-Type: application/x-font-ttf
Content-Length: 316664
Last-Modified: Fri, 22 Jul 2016 03:31:57 GMT
Expires: Thu, 31 Dec 2037 23:55:55 GMT
Cache-Control: max-age=315360000
Cache-Control: public
Accept-Ranges: bytes
Via: 1.1 vegur
Note that I ping the server directly without going through the CDN, the CDN then after invalidating all content should just forward whatever the server responds. The important line here is Server: nginx/1.10.0, which indicates that assets are served by nginx and not Rails. As a consequence, the rack-cors configurations do not apply.
The solution that worked for us is here: http://monksealsoftware.com/ruby-on-rails-cors-heroku-passenger-5-0-28/
It basically involved cloning and modifying the nginx config file for Passenger, which is not ideal since this copy needs to be maintained every time Passenger gets upgraded and the template changes.
===
Here's a summary:
Navigate to the root folder of your Rails project and make a copy of the nginx config template
cp $(passenger-config about resourcesdir)/templates/standalone/config.erb config/passenger_config.erb
Open config/passenger_config.erb and comment this line out
<%# include_passenger_internal_template('rails_asset_pipeline.erb', 8, false) %>
Add these configurations below the line mentioned above
### BEGIN your own configuration options ###
# This is a good place to put your own config
# options. Note that your options must not
# conflict with the ones Passenger already sets.
# Learn more at:
# https://www.phusionpassenger.com/library/config/standalone/intro.html#nginx-configuration-template
location ~ "^/assets/.+\.(woff|eot|svg|ttf|otf).*" {
error_page 490 = #static_asset_fonts;
error_page 491 = #dynamic_request;
recursive_error_pages on;
if (-f $request_filename) {
return 490;
}
if (!-f $request_filename) {
return 491;
}
}
# Rails asset pipeline support.
location ~ "^/assets/.+-([0-9a-f]{32}|[0-9a-f]{64})\..+" {
error_page 490 = #static_asset;
error_page 491 = #dynamic_request;
recursive_error_pages on;
if (-f $request_filename) {
return 490;
}
if (!-f $request_filename) {
return 491;
}
}
location #static_asset {
gzip_static on;
expires max;
add_header Cache-Control public;
add_header ETag "";
}
location #static_asset_fonts {
gzip_static on;
expires max;
add_header Cache-Control public;
add_header ETag "";
add_header 'Access-Control-Allow-Origin' '*';
add_header 'Access-Control-Allow-Methods' 'GET, HEAD, OPTIONS';
add_header 'Access-Control-Allow-Headers' '*';
add_header 'Access-Control-Max-Age' 3628800;
}
location #dynamic_request {
passenger_enabled on;
}
### END your own configuration options ###
Change the Procfile to include this custom config file
web: bundle exec passenger start -p $PORT --max-pool-size 2 --nginx-config-template ./config/passenger_config.erb
Then deploy...
===
If you know of a better solution, please put in the comments.
After implementing, the CURL command yielded the following response:
curl -H "Origin: https://tidyme-staging.com.au" -I http://tidyme-staging.com.au/assets/31907B_4_0-588bd4e720d4008295dcfb85ef36b233ee0817d7fe23c76a3a543ebba8e7c85a.ttf
HTTP/1.1 200 OK
Connection: keep-alive
Server: nginx/1.10.0
Date: Wed, 03 Aug 2016 01:43:48 GMT
Content-Type: application/x-font-ttf
Content-Length: 316664
Last-Modified: Fri, 22 Jul 2016 03:31:57 GMT
Expires: Thu, 31 Dec 2037 23:55:55 GMT
Cache-Control: max-age=315360000
Cache-Control: public
Access-Control-Allow-Origin: *
Access-Control-Allow-Methods: GET, HEAD, OPTIONS
Access-Control-Allow-Headers: *
Access-Control-Max-Age: 3628800
Accept-Ranges: bytes
Via: 1.1 vegur
As of version 5.0, Rails allows for setting custom HTTP Headers for assets and you don't have to use the rack-cors or font-assets gems. In order to set Access-Control-Allow-Origin for assets (including fonts), just add the following code to config/environments/production.rb:
config.public_file_server.headers = {
'Access-Control-Allow-Origin' => '*'
}
The header value could also be a specific domain, like the following:
config.public_file_server.headers = {
'Access-Control-Allow-Origin' => 'https://www.example.org'
}
This worked for my app and I didn't need to change any settings on Cloudfront.
I just had the same issue and managed to solve it.
You've correctly told Cloudfront to allow those headers, but you haven't added those headers to where Cloudfront gets the font. Yes, your origin headers are allowed, but Heroku isn't sending those headers with the font anyway.
To fix this, you'll need to get the proper CORS headers added to the font on Heroku. Luckily, this is pretty easy.
First, add the rack/cors gem to your project. https://github.com/cyu/rack-cors
Next, configure your Rack server to load and configure CORS for any assets it serves. Add the following after your application preloads in config.ru
require 'rack/cors'
use Rack::Cors do
allow do
origins '*'
resource '/cors',
:headers => :any,
:methods => [:post],
:credentials => true,
:max_age => 0
resource '*',
:headers => :any,
:methods => [:get, :post, :delete, :put, :patch, :options, :head],
:max_age => 0
end
end
This sets any resources returned from Heroku to have the proper CORS headers applied. You can restrict the application of headers depending on your file and security needs.
Once deployed, go into Cloudfront and begin an invalidation on anything that was previously giving you a CORS permission error. Now when Cloudfront loads a fresh copy from Heroku, it will have the correct headers, and Cloudfront will pass those headers on to the client as previously configured with your Origin permissions.
To make sure you're serving the proper headers from your server, you can use the following curl command to validate your headers:
curl -I -s -X GET -H "Origin: www.yoursite.com" http://www.yoursite.dev:5000/assets/fonts/myfont.svg
You should see the following headers returned:
Access-Control-Allow-Origin: www.yoursite.com
Access-Control-Allow-Methods: GET, POST, DELETE, PUT, PATCH, OPTIONS, HEAD
Access-Control-Max-Age: 0
Access-Control-Allow-Credentials: true
Here is a repo the demonstrates serving a custom font with Rails 5.2 that works on Heroku. It goes further and optimizes serving the fonts to be as fast as possible according to https://www.webpagetest.org/
https://github.com/nzoschke/edgecors
Asset Pipeline and SCSS
Place fonts in app/assets/fonts
Place the #font-face declaration in an scss file and use the font-url helper
From app/assets/stylesheets/welcome.scss:
#font-face {
font-family: 'Inconsolata';
src: font-url('Inconsolata-Regular.ttf') format('truetype');
font-weight: normal;
font-style: normal;
}
body {
font-family: "Inconsolata";
font-weight: bold;
}
Serve from CDN with CORS
I'm using CloudFront, added with the Heroku Edge addon.
If you're using your own CloudFront, make sure to configure it to forward the browser Origin header to your backend origin.
First configure a CDN prefix and default Cache-Control headers in production.rb:
Rails.application.configure do
# e.g. https://d1unsc88mkka3m.cloudfront.net
config.action_controller.asset_host = ENV["EDGE_URL"]
config.public_file_server.headers = {
'Cache-Control' => 'public, max-age=31536000'
}
end
If you try to access the font from the herokuapp.com URL to the CDN URL, you will get a CORS error in your browser:
Access to font at 'https://d1unsc88mkka3m.cloudfront.net/assets/Inconsolata-Regular.ttf' from origin 'https://edgecors.herokuapp.com' has been blocked by CORS policy:
No 'Access-Control-Allow-Origin' header is present on the requested resource. edgecors.herokuapp.com/
GET https://d1unsc88mkka3m.cloudfront.net/assets/Inconsolata-Regular.ttf net::ERR_FAILED
So configure CORS to allow access to the font from Heroku to the CDN URL:
module EdgeCors
class Application < Rails::Application
# Initialize configuration defaults for originally generated Rails version.
config.load_defaults 5.2
config.middleware.insert_after ActionDispatch::Static, Rack::Deflater
config.middleware.insert_before 0, Rack::Cors do
allow do
origins %w[
http://edgecors.herokuapp.com
https://edgecors.herokuapp.com
]
resource "*", headers: :any, methods: [:get, :post, :options]
end
end
end
end
Serve gzip Font Asset
The asset pipeline builds a .ttf.gz file but doesn't serve it. This monkey patch changes the asset pipeline gzip whitelist to a blacklist:
require 'action_dispatch/middleware/static'
ActionDispatch::FileHandler.class_eval do
private
def gzip_file_path(path)
return false if ['image/png', 'image/jpeg', 'image/gif'].include? content_type(path)
gzip_path = "#{path}.gz"
if File.exist?(File.join(#root, ::Rack::Utils.unescape_path(gzip_path)))
gzip_path
else
false
end
end
end
The ultimate result is a custom font file in app/assets/fonts served from a long-lived CloudFront cache.
Probably best is to use rack-cors gem. In the spirit of do-it-yourself and a follow-up to #GeekJock's answer. If one doesn't want to use rack-cors gem, this is poor man's CORS headers handling for a situation where for example we care only about static fonts assets (e.g. replacing the font_assets obsoleted gem).
Like in the other answer, you put in app config:
config.public_file_server.headers = {
'Access-Control-Allow-Origin' => '*'
}
To handle OPTIONS pre-flight requests, you can write a route matcher somewhere under /lib:
module FontAssetsConstraint
FONT_EXTENSIONS = %w[eot svg ttf otf woff woff2].freeze
module_function
def matches?(request)
extension = request.params["format"]
extension.present? && FONT_EXTENSIONS.include?(extension)
end
end
And then add a route definition config/routes.rb to catch those to reply:
Rails.application.routes.draw do
# Respond to pre-flight CSRF requests for font assets
if Rails.configuration.public_file_server.enabled &&
Rails.configuration.public_file_server.headers.include?("Access-Control-Allow-Origin")
constraints FontAssetsConstraint do
match "*path", via: :options, to: ->(hash) { [204, Rails.configuration.public_file_server.headers, []] }
end
end
Alternatively to writing a route matcher and a definition you can create your own middleware to catch fonts:
class AssetsOptionsResponder
TYPES = %w(eot svg ttf otf woff woff2).freeze
def initialize(app)
#app = app
end
def call(env)
if env["REQUEST_METHOD"] == "OPTIONS" && targeted?(env["PATH_INFO"])
[204, access_control_headers, []]
else
#app.call(env)
end
end
private
def targeted?(pathinfo)
return if pathinfo.blank?
TYPES.include? extension(pathinfo)
end
def extension(pathinfo)
pathinfo.split("?").first.split(".").last
end
def access_control_headers
Rails.configuration.public_file_server.headers
end
end
Then in app config or an initializer, you can add this middleware:
Rails.application.configure do
if defined?(ActionDispatch::Static) &&
Rails.configuration.public_file_server.enabled &&
Rails.configuration.public_file_server.headers.include?("Access-Control-Allow-Origin")
config.middleware.insert_before ActionDispatch::Static, AssetsOptionsResponder
end
end
# Gemfile
gem 'rack-cors'
# config/initializers/cors.rb
## very permissive
origins '*'
resource '*', headers: :any, methods: [:get]
## example of specifying only what's necessary
origins 'app.example.com'
resource '/packs/*', headers: :any, methods: [:get] # webpack
resource '/assets/*', headers: :any, methods: [:get] # asset pipeline
I'm trying to make X-Sendfile work for serving my heavy attachments with capistrano. I found that X-Sendfile is not working with symlinks. How could I handle the files inside a folder symlinked by Capistrano so?
my web server is apache2 + passenger
in my production.rb:
config.action_dispatch.x_sendfile_header = "X-Sendfile"
in my controller action:
filename = File.join([Rails.root, "private/videos", #lesson.link_video1 + ".mp4"])
response.headers["X-Sendfile"]= filename
send_file filename, :disposition => :inline, :stream => true, :x_sendfile => true
render nothing: true
my filesystem structure (where a "->" stands for "symlink" and indentation means subfolder):
/var/www/myproject
releases/
....
current/ -> /var/www/myproject/releases/xxxxxxxxxxxx
app/
public/
private/
videos/ -> /home/ftp_user/videos
my apache config
XSendFile on
XSendFilePath / #also tried /home/ftp_user/videos
My application is able to serve small files, but with big ones it gives a NoMemoryError(failed to allocate memory)
I think it's not using x-sendfile, because the behavior is the same if I don't use it.
Here are the response headers of the file i'm trying to serve
Accept-Ranges:bytes
Cache-Control:private
Connection:Keep-Alive
Content-Disposition:inline
Content-Range:bytes 0-1265/980720989
Content-Transfer-Encoding:binary
Content-Type:video/mp4
Date:Sat, 01 Mar 2014 13:24:19 GMT
ETag:"70b7da582d090774f6e42d4e44ae3ba5"
Keep-Alive:timeout=5, max=97
Server:Apache/2.4.6 (Ubuntu)
Status:200 OK
Transfer-Encoding:chunked
X-Content-Type-Options:nosniff
X-Frame-Options:SAMEORIGIN
X-Powered-By:Phusion Passenger 4.0.37
X-Request-Id:22ff0a30-c2fa-43fe-87c6-b9a5e7da12f2
X-Runtime:0.008150
X-UA-Compatible:chrome=1
X-XSS-Protection:1; mode=block
I really don't know how to debug it, if it's a x-sendfile issue or if I'm trying to do something impossible for the symlinks problem
EDIT:
Following the suggested answer in the accepted one, it "magically" started working!
I created a capistrano task this way:
task :storage_links do
on roles(:web), in: :sequence, wait: 2 do
#creo i link simbolici alle risorse
within "/var/www/my_application/current/private" do
execute :ln, "-nFs", "/home/ftp_user/videos"
end
end
end
I didn't manage to run it after finalize_update, so i run it after the restart, by hand.
And i corrected my apache configuration in this way:
XSendFilePath /var/www/my_application
(before i was pointing x-sendfile to the ftp folder)
In my response headers also now X-Sendfile is not appearing, and i got a 206 - partial content, but everything seems to work and apache is serving files in the right way (also very heavy files).
I know this can be a security issue, but i will try to point it to the last release of my application cause pointing it to the current symlink is not working.
Maybe I found a solution. How did you make your symlinks?
maybe you did ln -s, and it's not enough
Here they suggest using ln -nFs, so he recognizes it's a directory that you are linking in
Take 2 domains: www.domain.com and sub.domain.com. Each hosted on the same server at /home/www and /home/sub respectively and using a different php.ini file through vhost configuration within lighttpd.
fastcgi.server = ( ".php" =>
((
"bin-path" => "/usr/bin/php5-cgi -c /home/www/php.ini"
))
)
$HTTP["host"]=="sub.domain.com" {
fastcgi.server = ( ".php" =>
((
"bin-path" => "/usr/bin/php5-cgi -c /home/sub/php.ini"
))
)
}
Is it possible to enable www.domain.com/sub to serve the content of sub.domain.com while still using the appropriate php.ini?
The major difference in the php.ini files we are using is the include_path. Is there an alternative way to alter this through the server config by directory or alias? Or within a single php.ini?
The motivation for this is that we only have an SSL certificate for the main www domain, but wish to serve sub content via SSL on the primary domain path.
Using lighttpd on debian.
Don't know if it's still valid, but I followed this a few years ago... sadly I don't use lighttp where I work right now so can't verify it will still work.
http://www.cyberciti.biz/tips/custom-phpini-file-for-each-domain-user.html
We have an issue with TS min_inflex_len and delta indexes on our production servers
I have everything working in development mode on OSX but when we deploy via capistrano to our Ubuntu server running passenger / apache, both delta indexing seems to stop as well as min_inflex_len
We're deploying as ubuntu user which also runs apache. We had an issue yesterday with production folder not being created but we manually created and I can see a list of the delta files in there now.
I've followed the docs through..
I can see the delta flag set to true on record creation but when searching it doesn't find the record. Once we rebuild index (as ubuntu user) I can find record, but only with full string.
My sphinx.conf file is as follows:
production:
enable_star: 1
min_infix_len: 3
bin_path: "/usr/local/bin"
version: 2.0.5
mem_limit: 128M
searchd_log_file: "/var/log/searchd.log"
development:
min_infix_len: 3
bin_path: "/usr/local/bin"
version: 2.0.5
mem_limit: 128M
Rebuild, start and conf work fine and my production.conf file contains this:
index company_core
{
source = company_core_0
path = /var/www/html/ordering-main/releases/20110831095808/db/sphinx/production/company_core
charset_type = utf-8
min_infix_len = 1
enable_star = 1
}
I also have this in my production.rb env file:
ThinkingSphinx.deltas_enabled = true
ThinkingSphinx.updates_enabled = true
My searchd.log file only has this in:
[Wed Aug 31 09:40:04.437 2011] [ 5485] accepting connections
Nothing at all appears in apache error / access log
-- EDIT ---
define_index do
indexes :name
has created_at, updated_at
set_property :delta => true
end
Not sure if it's the cause, but the version values in your sphinx.yml are for the version of Sphinx, not Thinking Sphinx - so you may want to run indexer to double-check what that value should be (likely one of 0.9.9, 1.10-beta or 2.0.1-beta).
Also: on the server, in script/console production, can you share the full output of the following (not interested in the value returned, hence why I'm forcing it to be an empty string - it'll just get in the way otherwise):
Company.define_indexes && Company.index_delta; ''
``
delta not working on production server for passenger user, you have to give the write permission to your passenger user when creating index and write it to db/sphinx/production folder.
Or you can set two line in your nginx/conf/nginx.conf
passenger_user_switching off;
passenger_default_user root;
Example:
passenger_root /usr/local/lib/ruby/gems/1.9.1/gems/passenger-3.0.0;
passenger_ruby /usr/local/bin/ruby;
passenger_user_switching off;
passenger_default_user root;