user12096's questions

When this Ubuntu Hardy server comes up it is not always guaranteed that the dns settings are correctly, if the network cable was not connected during boot.

I would prefer a simple fast and fixed dns setup.

I do not have the knowledge of the packages, this is what somehow installed by itself (we always try to go simple and standard ways without changing much that comes out of the box):

$ dpkg -l|grep dns
ii  dnsutils                              1:9.4.2.dfsg.P2-2ubuntu0.5            Clients provided with BIND
ii  libavahi-compat-libdnssd1             0.6.22-2ubuntu4.1                     Avahi Apple Bonjour compatibility library
ii  libdns36                              1:9.4.2.dfsg.P2-2ubuntu0.5            DNS Shared Library used by BIND
ii  libnss-mdns                           0.10-3ubuntu2                         NSS module for Multicast DNS name resolution

This is the network config that I selected during install:

$ cat /etc/network/interfaces 
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).

# The loopback network interface
auto lo
iface lo inet loopback

# The primary network interface
auto eth0
iface eth0 inet static
 address 10.1.1.2
 netmask 255.255.255.0
 network 10.1.1.0
 broadcast 10.1.1.255
 gateway 10.1.1.1
 # dns-* options are implemented by the resolvconf package, if installed
 dns-nameservers 10.1.1.1
 dns-search localdomain

This somehow changes, but I don't want it to change:

$ cat /etc/resolv.conf 
search localdomain
nameserver 10.1.1.1

What would be the straightforward solution for a fixed dns setting here?

I assume I just have to uninstall a package and do some settings manually?

Please explain the steps for this running machine and also for a fresh Hardy install.

(It would be nice to get the same information for the latest LTS but I could also ask a different question for it)

Please clarify on the relation between xen, VirtualBox, Sun vm and Oracle vm

• Which products still exist after Oracle acquired Sun?
• which of the products that still exists are derived from xen?

We have some virtual machines on vmware server 2.x running on 64 bit hardware and quite happy with it.

As vmware server will no longer be offered we are thinking to migrate to ESXi, which seems is free. We will have to install the specialized network cards but that's a minor problem.

But once left alone with a quite silently discontinued product there is some resistance to vmware.

VirtualBox seems to work: http://blogs.oracle.com/virtualization/2010/06/migrating_from_vmware_to_virtu.html

What other free (of licencing cost) options are there? We have windows server 2003 32 bit VMs and also linux 32 and 64 bit VMs to migrate. So xen does not seem an option, which does not run microsoft OSes.

Edit: I accepted grub's answer, because I find many aspects of the questions covered. But its not easy to choose, because all answers are very good. Thank you all! I feel quite well informed now and your answers really helped a lot.

Fans and transformers which are inside the server case create magnetic and electric fields.

Electric fields can be easily shielded, but what about magnetic fields, they can only be shielded with high cost materials like mu metal http://en.wikipedia.org/wiki/Mu-metal

If a hard drive is installed too close to an intense transformer field, how is the magnetically stored information on the ferromagnetic surfaces of the disk kept safe? Even if drives are shielded, where are the limits?

Is there some technical investigation or recommendation from manufacturers about this?

(I never heard about something and never had any problem but I am interested in some facts. This is much preferred over what you believe or a habit you developed. Please try to give some solid infromation.)

I have built and repaired many servers and sometimes I did put the harddrive on top of the power supply.

Edit: This question is not about frequencies that could affect the drive via the power or data connectors of the drive, those are electronically decoupled and that's another question.

Edit 2: The wikipedia page states that the motor inside the drive is shielded with mu metal. It is obvious that manufactureres have to take care of this. This question is about such influences from outside the drive.

We experience a lot of traffic and server load on a web server.

All I can find out is majestic12 accessing pages all the time.

I wonder how I can prevent majestic12 from indexing the site

Do they respect any robots.txt entry and how do I write such an enty?

We have a windows server 2003 machine with rdp service listening on the standard port 3389.

For security reasons this port is not opened on the router, but we have freesshd service running and a remote admin can login via ssh and this port is forwarded to external port 33001 for the first external user. This works great.

Now we have another admin who wants to work remote (he uses a different windows account, and connects from a different location, where he wants to use a dekstop, but needs to work on the same server.)

So this is basically a ssh port forwarding question.

Will the other user be able to login at the same time forwarding the same port 3389 to his machine?

Please keep in mind that there will be a second tunnel, and this second tunnel will also use the local port 3389 on the windows server.

We have 2 flatrates in the house and have double bandwith theoretically. There is a local network in the house that connects everything.

But when I am alone I wonder how I can use both connections at the same time.

I want to build a solution where I can browse the web and page requests are spread between the 2 connections.

I imagine there are expensive routers who can split the traffic between 2 lines (Edit: which I do not want to buy or install). But is there a good way to do it with linux (Edit: which I prefer)?

The solution I am looking for will split the requests already for one page (multiple images, css files, javascrfipt files) between the two lines.

I use to have owned php scripts on a dedicated server by user "webroot".

It would be easier for coding and administration if the scripts were owned by the apache2 user, "www-data" on my system. Also feels more simple and clean.

There is no ftp on this box and there are no other users or sites.

Why not have the php scripts owned by www-data? If there is anything against it, what is the worst that can happen?

I have to configure an ubuntu hardy server network interface.

The service hoster told me that this is the network data for the machine:

IP Range:  111.111.200.74 to 111.111.200.78
Netmask:   255.255.255.248
Broadcast: 111.111.200.79
Gateway:   111.111.200.73
Subnet:    111.111.200.72/29

I am only using the first IP address. I will update the /etc/hosts file with 111.111.200.74, but I am still unsure how the /etc/network/interfaces file should be. This is my plan:

auto lo
iface lo inet loopback

auto eth0
iface eth0 inet static
  address   111.111.200.74
  netmask   255.255.255.248
  network   111.111.200.???
  broadcast 111.111.200.79
  gateway   111.111.200.73

As you can see I don't know how to build the network line. How would I calculate the data for the network line and what is the result?

(I changed the first 2 octets of the subnet, they are not "111.111" in the real setup.)

Edit: This is what I will use, constructed from the accepted answer below:

auto lo
iface lo inet loopback

auto eth0
iface eth0 inet static
  address   111.111.200.74
  netmask   255.255.255.248
  # the following line is optional/redundant, 
  # because address and netmask are given above:
  network   111.111.200.72
  broadcast 111.111.200.79
  gateway   111.111.200.73

Is there a way to check the result of a script with monit? For example a script returns 0 means ok, but 1 means failed. The idea is to call the script from monit on the local machine directly and interpret the result and send emails.

For example a script could do some hardware check.

My workaround is to call the script with cron and write the result to a file and check the file with monit.

I am looking for a soultion for trusted users to upload pdf files via html forms (with maybe php involved). This is quite a standard ubuntu linux server with apache 2.x and php 5.

I am wonderiung what are the benefits of the apache file upload module. There were no updates for some time, is it actively maintained? What are the advantages over traditional php upload with apache 2 without this module?

http://commons.apache.org/fileupload

I remember traditional php file upload is difficult with some pitfalls, will the apache file upload module improve the situation?

The solution I am looking for will be part of an existing website and be integrated into the admin web frontend. Things I am not considering are webdav, ssh, ftp, ftps, ftp over ssh. Should work with a browser and without installing special client software, so I am asking about a browser based upload without special client side requirements. I can request a modern browser like firefox >= 3.5 or modern webkit broser like chrome or safari from the users.

I have a HP laserjet shared on a windows 2003 server as //w3kserver/hp

I want to print from a linux client that has the cups system installed. When I add the printer in the cups web frontend on the linux client I need to provid an url of the printer that is shared on the windows server. How to construct that url for cups?

How do I extend a selection in nano and pico with the keyboard, so I can move and copy blocks?

Sorry, this is a question I would normally find out by myself but I have searched a lot in different occasions now without success.

I am installing ubuntu server in a vmware vm. In the expert installer there is this question (I guess this is the same in the debian installer):

Please enter a space delimited list of virtual consoles you use

and the standard setting is:

/dev/tty[1-6]

I wonder if I can reduce this to maybe 1-2. I am not sure what it exactly means. I will only use one "physical" console for install one time and maybe later for emergencies, but normally only ssh remote login consoles.

Maybe this is related to this faq entry:

http://www.faqs.org/docs/Linux-HOWTO/Keyboard-and-Console-HOWTO.html#ss7.1

So my question is: what does it change if I reduce this to 1-2? I am wondering on each install.

EDIT: I selected this for my install:

/dev/tty[1-1]

But later on the running instance I got 6 consoles with Alt-1 .. Alt-6 on the vmware console nonetheless. Strange. There are 6 instances of getty showing up in the processlist.

EDIT 2: I should have added that this was the Ubuntu 10.04 server install Disk (i386) "Lucid Lynx" beta 2

Where is the bare cygwin package list located and how do I manipulate it programmatically or from a shell or with a different method than the gui? I know the gui (setup.exe), and I'd love to go one or more levels deeper.

I can retrieve a list of selected/installed packages ( Cygwin package management ), but how do I write it back or to a different machine?

What I have in mind is when I install a new windows I would like to start with my package list in text form, an apply or inject it somehow to the new system.

Where is it? In the registry? In a binary file? in a local database?

Or has anybody done this, is there a tool, a tutorial? The essence of what I want is to manipulate the selected package list with something else than the gui. It is ok for me to use the gui for the setup process. So I could imagein manipulating the package List and then run setup.exe and just click through it.

Note: I do not want to manipulate the list of already installed packages but of packages that "should be installed". But if htis is not possible, maybe there is some workaround. E.g add an outdated version as installed and the installer will then install the new version.

I am working on an apache2 reverse proxy vhost. When the server behind apache is down, the first request to apache shows the error page of course.

But at subsequent requests it seems apache delays for some time before asking the backend server again. During all this time (which is short but in development I don't want a delay at all) only the apache error page is shown to the browser, although the backend server is already up.

Where is this setting in apache, what is this behaviour, and how can I set the delay time to zero?

Edit: I am not trying to change the timeout for a single request. I want to change the blocking time. It is my experience that apache blocks further requests for a certain time before asking a backend server again that has failed once.

Edit2: This is what apache delivers:

Service Temporarily Unavailable

The server is temporarily unable to service your request due to maintenance downtime or capacity problems. Please try again later.

After hitting Ctrl-R in firefox for 60 seconds the page finally appears.

I have a mountpoint /mnt/v1 and a user bak in group bak. I did this:

sudo chown bak:bak /mnt/v1
sudo chmod 775 /mnt/v1

Now I want to mount an external fat32 drive with partition /dev/sdb1 to this directory. I get the message "only root can do that", so I found out I need to change fstab and added this line:

/dev/sdb1       /mnt/v1     vfat        noauto,user           0      2

But still I get the nasty message "only root can do this" when I mount like this:

bak$ mount -t vfat /dev/sdb1 /mnt/v1

But when I mount as root the whole drive get permissions of root which I don't want.

Is there some action necessary so the new fstab entry get's read into the kernel? How can I achieve what I want?

I want to add a user to the linux system from a script, but I don't want to invent or care for the password. This should be done automatically.

The goal is to generate ssh-keys and this user needs only to login remotely. Or alternatively this user will be use from a sudo user via sudo su - thatuser.

I want to save the hassle inventing and typing some secure password every time I create such a user, and also do this from scripts.

No one should be able to login with password as this user, so my idea is he gets a good random password but nobody knows it.

I could write a script that generates something ramdom but mabye there is something built in? Or just deactivate the password (so that password login is not possible, but ssh login with keys and sudo su - thatuser works fine.

Edit: There are already some answers, great, but I am still not sure how to do it. How would the script look like?

It should behave like adduser and create everything standard (like std. home dir, skeleton copied, group with the same name)

Edit2: In the end and with your the help I found a solution that works and I want to share it. This is a script that I call "adduser-nopasswd" and I put it into /usr/local/sbin (is this a good place?) and it can be executed only by root. It takes one argument, which is the name for the new group and user at the same time:

#!/bin/sh -e
# the -e makes the script exit immediateley if one command fails
NAME=$1
groupadd $NAME
useradd --create-home -d /home/$NAME --shell /bin/bash -g $NAME $NAME

Any comments on this function?

I have an /etc/init.d/iptables start|stop|restart script on different ubuntu servers (which is a normal shell script)

For each new service I have to edit and insert a line to open a port. This leads to many different versions of the init.d script on different machines.

Is it possible to automatically include let's say all files in /etc/iptables/include.d/ ?

The target is that there should only a line in the start function of /etc/init.d/iptables like

include /etc/iptables/include.d/*

And after an additional file in /etc/iptables/include.d/ I'd simply say

/etc/init.d/iptables restart

Edit: As Saurabh pointed out this can lead to problems when commands need a certain order. An advanced setup could have different directories like:

/etc/iptables/include01.d/
/etc/iptables/include02.d/
/etc/iptables/include03.d/

and including them like this:

    include /etc/iptables/include01.d/*
    ... maybe some code goes here in the main file...
    include /etc/iptables/include02.d/*
    include /etc/iptables/include03.d/*

I want to enable the simple php mail() function on an Ubuntu server. This will only be used for sending administration stuff and logs. This is all the mail functionality that I want on this machine.

1) Can you name 1 or 2 simple to setup standard ubuntu packages that I can use for this? Should take few resources and few dependencies. Which are the most common packages for this purpose and what are the pros and cons for what I want to do with it? (This machine wll never receive mail.)

2) After I have chosen a mailer package, how will I test it from the command line? Is it correct I will have to allow outgoing connections to remote port 25?

3) Are there any steps to make sure php can use it now for the mail() function?

I want this as a recipe for future server installations I can follow, so it would be interesting to know what can go wrong and how to deal with it.

Edit: I installed sendmail in the end, and this removed exim4. Php mail() worked after that. So for me sendmail seems easier than exim4. The formerly installed php5 with exim4 did not send mails.