Our mailserver is a Ubuntu Server box with a Postfix/Dovecot combination. We enabled starttls with a correct certificate and everything is working ok.
A lot of our users are using Thunderbird for retrieving mail with IMAP. Thunderbird tries to guess the configuration of our server (or autodetect it), but that always fails, resulting in that our users need to give the settings manually.
What are the steps we should take to 'help' thunderbird and enabling auto-configuration?
I have set up nginx with SSL. Everything works perfectly, with online tools giving the domain a good score.
Now I am wondering about one particular nginx configuration option; ssl_dhparam. Should I generate and set these parameters? Does it have any influence on security or computational load of SSL?
How can I prevent Nginx from buffering the output of my uwsgi app? For my comet style application I'm using long polling and the requests are now buffered.
I tried to reduce the size of the buffers, but I'm not allowed to put uwsgi_buffer_size and uwsgi_buffers on 0. Also uwsgi_max_temp_file_size does not work (eventhough the manual suggests that).
How can I do this?
Possible Duplicate:
How to configure a shortcut for an SSH connection through a SSH tunnel
Multiple networks I can connect to have the following set up. There is a single node that can be accessed with ssh from the internet, and from there you can ssh to the inside network.
Is there a way, that on my ubuntu box (no sudo rights), I somehow can configure that the command ssh internal.host is first setting up an connection to external.host and from there to internal.host? In the ideal situation it would be something very transparent, so that I even would be able to do scp myfile internal.host:
All authentication is done with certificates.
Btw. I am aware of ssh tunnels. I'm only searching something more transparent, I don't want to have to always manually open an ssh tunnel to reach a host.
I would like to find a tool to test the network bandwidth between 2 servers. Also statistics about latency and package reliability would be nice.
I have shell access to both servers, but no root access. I can run open some high numbered port. The machines are running CentOS.
I have tried to test speeds with scp, but I think what I get to see is the disk write speeds instead of the network speed.
For my webpage at my university I can place files in a directory that can be served by apache. Also at least php is supported.
What I am wondering is, what apache modules are available? E.g. I want to know or I can use wsgi or mod_python.
How can I discover or this modules are enabled? I have access to: all files in my directory (including .htaccess) and the access + error logs. I don't have access to the webserver itself. Is there some easy way I can display all modules in use or other config variables?
To be able to use ec2-consistent-snapshot on my EC2 servers I have decided to make my partition for my database data XFS
What are the best XFS configuration options I can use for these disk? The disk will be around 5-10 Gb big, containing 1 Gb of database. I am looking for both the best options when creating the filesystem as the mounting in fstab.
The operating system I use is Ubuntu Server 10.10
I am have a small website that we want to deploy to Amazon EC2. The reason is that it is cheaper than most other VPS systems, and we want to have completely custom configuration so a shared host is not a possibility. Also, the option to scale to a bigger server with the same images is one of the reasons of going for EC2.
Because we start with a t1.micro instance we will use EBS booted instances (x64_64). Our operating system of choice is Ubuntu Server. We don't compile any packages ourselve, sticking to stable packages from the repositories.
All our webserver data is stored on another EBS data volume.
Now the question: How can we best deploy our webserver, while keeping things up to data.
The options I have considered so far:
What is the best way to do it? Are there any good options I missed, or are there any other aspects to my options that I didn't consider.
In my Nginx setup I would like to password protect some website, except when I access it from my ip at home. How do I set this up?
Also, except for getting access to my home pc, is this safe?
I have a Ubuntu Lucid server with Nginx (0.7.65) and php5-fpm (package by Brian Mercer)
All my configuration is default and my site config is looking like this (domains replaced):
server {
listen 80;
server_name sub.example.com;
access_log /data/log/www/sub.example.com/access.log;
error_log /data/log/www/sub.example.com/error.log;
location / {
root /data/www_data/sub.example.com/public;
index index.php;
}
location ~ \.php$ {
fastcgi_param SCRIPT_FILENAME /data/www_data/sub.example.com/public$fastcgi_script_name;
fastcgi_pass 127.0.0.1:9000;
include /etc/nginx/fastcgi_params;
}
}
And I have an index.php present with <?php phpinfo(); ?> in it.
Now when I load sub.example.com/ in my Chrome webbrowser I get to see the phpinfo. The strange thing is, in my access log multiple entries are written for the index.php hit. For the favico (which does not exist) only one hit. See the log below.
130.233.178.70 - - [23/Sep/2010:05:00:27 +0000] "GET / HTTP/1.1" 200 10287 "-" "Mozilla/5.0 (X11; U; Linux x86_64; en-US) AppleWebKit/534.3 (KHTML, like Gecko) Chrome/6.0.472.62 Safari/534.3"
130.233.178.70 - - [23/Sep/2010:05:00:27 +0000] "GET /index.php?=PHPE9568F34-D428-11d2-A769-00AA001ACF42 HTTP/1.1" 200 2536 "http://sub.example.com/" "Mozilla/5.0 (X11; U; Linux x86_64; en-US) AppleWebKit/534.3 (KHTML, like Gecko) Chrome/6.0.472.62 Safari/534.3"
130.233.178.70 - - [23/Sep/2010:05:00:27 +0000] "GET /index.php?=SUHO8567F54-D428-14d2-A769-00DA302A5F18 HTTP/1.1" 200 2825 "http://sub.example.com/" "Mozilla/5.0 (X11; U; Linux x86_64; en-US) AppleWebKit/534.3 (KHTML, like Gecko) Chrome/6.0.472.62 Safari/534.3"
130.233.178.70 - - [23/Sep/2010:05:00:27 +0000] "GET /index.php?=PHPE9568F35-D428-11d2-A769-00AA001ACF42 HTTP/1.1" 200 2158 "http://sub.example.com/" "Mozilla/5.0 (X11; U; Linux x86_64; en-US) AppleWebKit/534.3 (KHTML, like Gecko) Chrome/6.0.472.62 Safari/534.3"
130.233.178.70 - - [23/Sep/2010:05:00:27 +0000] "GET /favicon.ico HTTP/1.1" 404 143 "-" "Mozilla/5.0 (X11; U; Linux x86_64; en-US) AppleWebKit/534.3 (KHTML, like Gecko) Chrome/6.0.472.62 Safari/534.3"
What is causing this triple hit on index.php?
I have seen this answer for growing EBS volumes, but I would like to shrink one.
The default Ubuntu Server images are 15 GB while I really need only 2 GB max (I use a different volume for data). Is there a way to shrink the size of the volume?
I have the following directory structure for my websites:
/var/www/sitename/subdomain/(public|log)
e.g /var/www/stackoverflow.com/careers/public/index.html
Can I make a single generic nginx configuration to do this? So that every domain is mapped to the right directory? I would not like to edit my nginx configuration for every website I add.
The root domain can always be mapped to the www subdomain.
I am aware of using lsof for checking the files currently accessed by a process. Does there exist a way to see all files that an application opens in its lifetime?
Who knows where the gpg key for dotdeb.org repository (for debian lamp packages) can be found?
At this moment I am preparing a virtual server to host a web application which needs php5.3
The virtual server base image is always Hardy (8.04 LTS). There is no php5.3 until the upcoming release in a few days: Lucid (9.04 LTS).
I am seeing to options: - waiting until the final version is released and then start preparing this server - Now upgrading to the beta (do-release-upgrade --devel-release) and when the final release has come upgrading to that
For time constraints I would prefer the second option. I only can't find whether it will be easy to upgrade from a beta to the 'clean' final release. Is this possible in an easy way. Will it have any drawback for security or will there be any traces left of it being ever a beta release?
Note: the server will not go into production before the LTS is really installed.
For a kind of Internet Café we want to do the following:
How to set this up?
(Later we maybe want to add a warning 5 minutes before shutting of)
I need to set up an OpenVPN network with two protected segments.
1) A developer network where all developers can reach each other machines. 2) A support network with two roles, supporter and client. Preferably clients can not initiate connections to the supporter, but the support can initiate connections to clients.
The easy way to set up this would be with two separate OpenVPN configurations, 2 key infrastructures etc.
However, I think this should be also possible with only one OpenVPN configuration, allowing people who are both supporter and developer to use only one certificate for both purposes.
How should I set this up?
(OpenVPN version 2.0.x, openSuse 11.1)
We are running computing jobs with GridEngine. Every jobs returns 3 different times:
What are the differences between these three? Which of these three is most suitable to compare the performance of two applications/scripts
I need to gather a list of files and put this in a textfile. However, the files should not include their extension.
The command I have now without removing file extensions:
ls -1 /a/dir/*/dir/* > textfile
All selected files will have a an extension of .[a-z]{3}
(the reason I need this is because I need a list of unique base names, and unique will solve that after this step)
My situation: I want to remotely debug my c++ application with gdbserver and eclipse. Eclipse is running on my local workstation (in the 192.168.0.0 network) and gdbserver is running on a clustermachine in the 10.0.0.0 network. I have ssh access to the clustermachine via a machine that is in both networks.
How can I set up a tunnel so that my eclipse can access the gdbserver? I don't have administrator rights/root/sudo access.
All the machines are linux/Unix with openssh.