Hello,Simon
Thanks for your help.But after I got the ip and edit /etc/hosts file as you
said,when I come to connect to my app-server, there comes a page shows"404 not
found / nginx/1.5.0".
At the same time ,when I run my Hilary with "node app.js |
node_modules/.bin/bunyan", There are 3 errors
1.ERROR: oae-modules/2574 on ip-10-120-51-87: Error initializing module oae-doc
err: { "code": 404, "msg": "No documentation for this module
was found" }
2.ERROR: system/2574 on ip-10-120-51-87: Error initializing server. err: {
"code": 404, "msg": "No documentation for this module was found"
}
3.[2013-05-18T10:43:08.537Z] ERROR: oae-cleaner/2506 on ip-10-120-51-87: Could
not list the files. (directory=/tmp/oae) Error: ENOENT, readdir '/tmp/oae'
What should I do ?
Thanks a lot !
Wilson
------------------ Original ------------------
From: "Simon Gaeremynck"<gaeremyn...@gmail.com>;
Date: Sat, May 18, 2013 04:12 PM
To: "Wilsonlee"<281579...@qq.com>;
Cc: "oae-dev"<oae-dev@collab.sakaiproject.org>;
Subject: Re: [oae-dev] About Nginx configuration
Hi Wilson,
The way we have it set up, there are (at least) 3 "servers".
1. The admin UI server
ex: http://admin.oae.com
2. An etherpad server (when fully scaled out, you'll have more)
ex: http://0.etherpad.oae.com
3. A catch_all server for tenants
ex: http://tenant1.oae.com
These can all be configured to run in 1 nginx machine though.
I'm assuming you're deploying this on AWS as a testcase and not as a full
production cluster.
Given that assumption we'll fake DNS resolution by modifying the local
/etc/hosts file.
Nginx.
---------
1. The Admin UI server [1]
Handles all the requests to the global admin UI.
We need to give this a server_name so nginx knows that it should "listen" on
that address.
server_name admin.oae.com;
2. An etherpad server [2]
Etherpad has a certain lack of clustering/gracefull failover. To alleviate that
a bit,
the application will shard requests over a cluster of etherpad servers. It
assumes there are
a pool a n servers available at [0-n].etherpad.oae.com. (the hostnames can be
configured in config.js)
In deployment there will obviously only be 1.
server_name 0.etherpad.oae.com
3. A catch_all server for tenants [3]
All requests that cannot be routed to the admin ui or an etherpad server are
assumed to
be requests to tenant servers. This is why this server block needs no
server_name.
nginx will automatically proxy the requests to the Hilary server.
DNS.
------
Unless you're able to configure 3 DNS record (or a wildcard record) the best
way to verify your deployment is to configure
your /etc/hosts file
You will need a publicly reachable IP address, so I *think* you'll need to
configure an elastic load balancer which points to
your EC2 instance. Assuming that gives you an IP of 98.98.98.98, you can
configure your /etc/hosts file with:
98.98.98.98 admin.oae.com
98.98.98.98 0.etherpad.oae.com
98.98.98.98 tenant1.oae.com
Make sure you flush your DNS cache (on OS X that's: sudo dscacheutil
-flushcache) before trying again.
Let us know if that helps,
Simon
[1] https://github.com/sakaiproject/3akai-ux/blob/newframework/nginx.conf#L30
[2] https://github.com/sakaiproject/3akai-ux/blob/newframework/nginx.conf#L320
[3] https://github.com/sakaiproject/3akai-ux/blob/newframework/nginx.conf#L148
On 18 May 2013, at 03:58, Wilsonlee <281579...@qq.com> wrote:
> Hello everyone,
>
> When I connect to my app-server, the page appears the "Welcome to nginx!". So
> I think if there is sth wrong with my nginx conf(especially the server_name
> property).I'm deploying the server on AWS and could anyone give me an example
> about how to deal with that property??
>
> Thanks a lot!!
> _______________________________________________
> oae-dev mailing list
> oae-dev@collab.sakaiproject.org
> http://collab.sakaiproject.org/mailman/listinfo/oae-dev_______________________________________________
oae-dev mailing list
oae-dev@collab.sakaiproject.org
http://collab.sakaiproject.org/mailman/listinfo/oae-dev
