Expanding the size of a Windows Guest VM hard drive on Mac-hosted Virtualbox

Assume that you are using OSX as your “host” OS and Windows 7 as your “guest” OS. You accidentally create a virtual disk that is too small in Windows. Now you’ve run out of space and you need to expand the size of the disk. Virtualbox to the rescue! It allows you to do this, but only from the command line. There is an excellent tutorial on this over here: http://hajuria.blogspot.com/2013/12/how-to-increase-virtualbox-disk-size-in.html .

Continue reading

Powershell Tuesday

Here’s a Powershell one-liner (or two) for your Tuesday.

1. Imagine you have a file of NetIDs, one per line, and you would like to generate a CSV file of first names, last names, etc. from the file of NetIDs. Here’s how you do it:

get-content C:\Users\bjchapm\Documents\badnetids.txt | get-aduser -Properties Name,Givenname,Surname | select-Object Name,GivenName,Surname

2. Really quick-and-dirty (and not very correct) dump of all of the groups that a particular set of your users belong to:

Import-Module ActiveDirectory
$Users = get-ADUser -Filter {(objectCategory -eq "User") -and (Description -eq "Student:School of Law")} -Properties MemberOf,Name,FullName |Select-Object MemberOf,Name, FullName
foreach ($user in $Users) {
 $user.name, $user.fullname, $user.MemberOf
 }

This will print out (to the console) each user name and the groups to which the user belongs. However, it’s clearly not the right way to do it… but, it worked for me.

Powershell is actually fun. 🙂

In Hell, You Have To Use MS Word Mail Merge

I’m starting to get old. How do I know this? Well, one way is that I remember with fondness mail merges in WordPerfect. Ah, but we don’t use WordPerfect any longer. We use Word. Word has a perverse mail merge system. I have no idea how office admins (whom I assume are the target market for this feature), learn how to use this system efficiently. My hat is off to them.

Some years ago, in an effort to save money, I inherited the task of creating an electronic face book of our students. It shows some basic information:

  1. A photo (if we have one)
  2. The student’s name
  3. Their undergraduate institution
  4. Their major(s)
  5. Their graduate institution, if any, along with graduate degree earned, as well as area of graduate study
  6. Their permanent city and state

I have produced this both in gridded form and row form for some time.

To generate these, I’ve used the fantastic ReportLab Python PDF library, and I’ve also used Pentaho’s Report Designer. The backend data is stored in a spreadsheet and/or a database table. The issue, of course, is that I’ve never turned this service back into a turn-key offering that could be handed off to someone else. This is the curse of the small shop (or the unskilled coder). You have enough time to write 60% of what you need to write – enough to get the job done and enough to get some kudos on the first iteration of the product.

However, you now own what was essentially an administrative task and one that would have been done with glue and scissors 20 years ago and would not have required IT intervention. So, I started looking at MS mail merge, because I thought, how hard could it be? Right? Perhaps that would be a way to hand off this task to someone else. Geez.

Here’s a screen capture of what the merge codes might look like to load up one entry:

Mail merge for a face book.

Really? This is what is required for an admin to come up with a reasonably basic entry that includes a picture based on the value in a field and also deals with missing data reasonably well. There are lots of neat nested expressions. Whitespace is significant to printing, as well, so you’ll need to triple check that. Note that I haven’t yet dared to tackle blank lines generated by the absence of grad-school data. It gets better… If you load up a label template with all of this code, you will not be able to see the bottom parts of it. You will also need to deal with a cryptic wizard or two along the way.

Here are some references that helped me get this far:

As for me, I think I’ll go back to updating the Python script that creates these. Once again, I tip my hat to the people who have to use MS mail merge every day. Perhaps this set of links will be helpful – also, of course, I could just be a complete ignoramus when it comes to this stuff. Any tips are most appreciated. Feel free to share in the comments.

Adding the Shared Folders MMC snap-in to Windows XP

Shows how to customize the Windows XP MMC to add support for managing Shared folders. This is really old-hat, but I thought I would write it up anyway, as it’s something that comes up periodically. Note that these instructions are written from the perspective of using a Windows XP SP3 workstation to manage a Windows 2003 R2 server – this may or may not be relevant to new versions of Microsoft’s operating systems.

 

Start up MMC (Microsoft Management Console

media_1297697641387.png

Start > Run > “MMC” > OK

Customize MMC by adding Shared Folder management

media_1297697722187.png

Click on File
Click on Add/Remove Snap-in

Select appropriate snap-in

media_1297697843320.png

Make sure the Standalone tab is selected.
Click Add.
Select Shared Folders
Click Add

Choosing options for console

media_1297697967700.png

When you click “Add”, it will display this options dialog.
In this case, I have chosen to manage a remote server and not this computer.
Once you have chosen the options you want, click “Finish”, then “Close”, then “OK”

Verifying functionality

media_1297698154882.png

Verify that you see the resources that you expected to see.
This console gives you the ability to manage Shares, Sessions, and Open Files on the server.

Checking share permissions

media_1297698260548.png

Choose Shares under Shared Folders.
Right click on the share that you want to check.
Click on Properties.
Review share permissions.

Customizing the MMC

media_1297699112384.png

The original idea of the Microsoft Management Console was that people would use the snap-in architecture to customize and create consoles of just the tools that they used.
The screenshot is one example of this. The Shared Folders snap-in has been inserted twice, pointing at two different servers. Also, if you click File > Save As, you can save the console for re-use. As you can see above, I’ve saved this console as “BJCShareMan”. It’s accessibile under Start > All Programs > Administrative Tools > BJCShareMan.msc.

What is that device on my network?

It’s that time of year when I start poking around our network trying to update the various records that we keep of the various devices on our network. Unfortunately, we do not use any fancy network tools that automatically map our local network service to determine what machines are on the network. However, it’s not difficult to figure out at least a little bit of information and these are useful utilities to be familiar with. Here’s the process:

  1. Ping an address on your network.
  2. If you get a response, use the arp utility to find the mac address of the device that responded.
  3. Paste the first six digits of the mac address into a web page that will help you look up the manufacturer of the network card.

Continue reading

Kerberos Single Sign On and LDAP Authorization to Apache

Introduction

We have various web servers running Apache on CentOS 5.5. We have approximately 200 users who authenticate to the University domain every morning when they sit down at their Windows computers  and log-in. It would be great if we could “pass-through” that authentication to the Apache web servers so that our users do not have to log in again. Second, we want to check that they are authorized to access various resources on the web server. The first part, SSO authentication, is handled via mod_auth_kerb; the second part, authorization to use particular resources, is handled via mod_authnz_ldap. Mod_authnz_ldap is bundled with Apache 2.2. We will have to build a new version of mod_auth_kerb from source to get a feature that we need to make this work. While this has been documented in a number of places on the web, it took me a long time to get all of the pieces to work together, so I’m going to document our approach.

I’ll start with a list of some of the resources that I consulted while getting this to work:

My notes should be considered as a supplement to their work, and not a replacement for it. I am not an expert in this area by any possible stretch of the imagination.

Initial Configuration

I configured a Windows XP SP3 machine and added it to our domain. I also configured a CentOS 5 server . I did a yum groupinstall of “Web Server” and “Windows File Server”. I also installed the following packages on the CentOS box:

  • krb5-libs-1.6.1-36.el5_5.4
  • krb5-workstation-1.6.1-36.el5_5.4
  • krb5-devel-1.6.1-36.el5_5.4
  • pam_krb5-2.2.14-15
  • krb5-server-1.6.1-36.el5_5.4
  • mod_auth_kerb-5.1-3.el5 – CAVEAT: mod_auth_kerb 5.1 does not work in our environment, as discussed below. You can remove the mod_auth_kerb rpm. We will replace it with a mod_auth_kerb built from source.

On the Windows workstation, I installed:

  • Windows Server 2003 Administration Tools Pack
  • Windows Resource Kit Tools for Windows 2003.

Our Windows servers are running Windows 2003. I am a sub-admin on our Windows domain.

Configuration files

On the CentOS box, I created a file /etc/krb5.conf that looks like this:

[libdefaults]
    default_realm = EU.EMORY.EDU
[domain_realm]
    eu.emory.edu = EU.EMORY.EDU
[realms]
  EU.EMORY.EDU = {
           default_domain = eu.emory.edu
           kdc = DOMAINCONTROLLER.EU.EMORY.EDU:88
           admin_server = DOMAINCONTROLLER.EU.EMORY.EDU
   }

This works for us. Make sure that Kerberos is working from the Linux machine before you do anything else – make sure that you can do a “kinit yourname@YOURKERBEROSREALM” to test that your Linux box is talking to the Windows Kerberos server. The command “klist” should show that your account received an initial ticket, all as shown in step five of the Groims article. Be aware that Kerberos realms are always typed in UPPER case letters.

Working on the Windows client

As is discussed in the articles referenced above, it is necessary to connect to the Windows domain for the next part. Every kerberized host (each Apache server) must have a corresponding service principal in Active Directory. This is a two-step process. You create a user in AD, and then you use ktpass from the Windows machine to map the AD user to the HTTP service and create a keytab file that will be installed on the kerberized host.

This can be done from a domain member workstation; you do not need to be logged into the domain controller for this. I followed the instructions in Scott Lowe’s blog, referenced above. It is important to follow his advice to create a user account and not a computer account. Each Apache server needs a user account, as mentioned above. You may be tempted to create a computer account. It doesn’t seem to work. Create a user account instead.

I am a domain sub-admin within our organization. I do not have rights to create users in the main “People” container for our Active Directory tree. I do have the ability to create users within a User OU that is part of our local OU and that’s what I did. There are a couple of gotchas that might bite the novice (like me):

  • Follow the advice to create a user for the service account. Mr. Lowe’s naming scheme makes sense. We created a user called “http-els4182” to be the service principal for the HTTP server running on els4182.
  • Create the account using the ktpass command as documented in the resources above.
    • ktpass -princ HTTP/els4182.law.emory.edu@EU.EMORY.EDU -mapuser http-els4182 -crypto rc4-hmac-nt -ptype KRB5_NT_SRV_HST -pass CHOOSEPASSWORD -out http-els4182.keytab
  • Note that Mr. Lowe and Mr. Groims use different types of encryption. The method suggested by Mr. Groims is supposed to be better. I am not a security guru, but you should probably use rc4-hmac-nt if possible.
  • Pay attention to the “-out” switch. It will fail silently if you point it at a directory that doesn’t exist. It’s probably fine to create the file in the current directory, but if you don’t like that idea, point it to your desktop or similar.

At this point, you should be able to check that the service principal was properly mapped to the user account by doing this:

  • cd c:program filessupport tools
  • setspn.exe -L ADUSERNAME

In our example above, you would type setspn.exe -L http-els4182. You’ll get back something like this:

Image of screen capture of setspn

This shows that the service principal has been correctly configured.

Working on the Linux Server

For the second half of this exercise, you will need a newer version of the mod_auth_kerb than the one that is available in the repositories for CentOS. The reason is that the 5.1 version of mod_auth_kerb passes through username@REALM as the username. That doesn’t match our user names. As a result, Kerberos will succeed and the user will be authenticated; however, authorization will always fail. As discussed in the comments to this article, that was fixed in version 5.4 of mod_auth_kerb and a new option, KerbLocalUserMapping was added to mod_auth_kerb. It strips the REALM information from the username, allowing LDAP authorization to work.

Here’s how to take care of installing mod_auth_kerb:

  • Make sure that you have httpd-devel installed: yum install httpd-devel
  • Download the source to the latest version of mod_auth_kerb. On the server, do something like:
  • Uncompress the package: tar xvfz mod_auth_kerb-5.4.tar.gz
  • cd mod_auth_kerb-5.4
  • I added /usr/sbin/ to my path so that ./configure would be able to find apxs and I turned off krb4 support:
    • export PATH=$PATH:/usr/sbin
    • ./configure –with-krb4=no
    • make
    • (as root) make install
  • Stop and start the httpd server and verify that there are no errors.

Next, I created a new configuration file for Apache, located in /etc/httpd/conf.d/ . I named the file auth_kerb.conf; however, the naming is up to you so long as it ends in .conf. Here are the contents of that file. You will need to edit it as appropriate:

Alias /private/ /var/www/html/private/

<Location /private/>
Options Indexes
SSLRequireSSL
AuthType Kerberos
  AuthName "Emory Law Kerberos Login"
  KrbMethodNegotiate On
  KrbMethodK5Passwd On
  KrbAuthRealms EU.EMORY.EDU EMORY.EDU
  KrbServiceName HTTP
  Krb5KeyTab /etc/httpd/conf/els4182.keytab
  KrbVerifyKDC off
  KerbLocalUserMapping on
  require valid-user
</Location>

Refer to some of the references cited above for the rationale for the options. Basically, you’re configuring this location to require SSL and you’re telling Kerberos to fall back to basic authentication if the “negotiate” method of authentication fails. What this means is that, if you’re using Firefox on a Mac (and you aren’t using Kerberos), you’re going to see a standard basic authentication prompt asking you to enter your username and password. I have added KrbVerifyKDC off, because our Kerberos server uses self-signed certificates. I’ve added the critical KerbLocalUserMapping on section to make sure that Kerberos passes the username only into Apache rather than the username@REALM. This is important if you’re doing additional authorization via LDAP as discussed below. Finally, “require valid-user” means that anyone with an Active Directory account can access this portion of the web-site. We’ll limit that access via LDAP in a minute. At this point, you need to get the keytab file that you created on the Windows client up to the Linux host using scp or some other method. I chose to put the keytab in /etc/httpd/conf/els4182.keytab as suggested by Lowe. Once the keytab is in the right location, change its ownership to the apache user and make it read-only to that user:

# chown apache.apache /etc/httpd/conf/els4182.keytab
# chmod 400 /etc/httpd/conf/els4182.keytab

As suggested by Groims, test using kvno and klist – although I used a slightly different klist command:

# kvno HTTP/els4182.law.emory.edu@EU.EMORY.EDU
# klist -ke /etc/httpd/conf/els4182.keytab

By using “-k”, you can see the keys in the keytab file.

At this point, you should be able to fire up Apache and check access to the directory via Kerberos from a properly configured Windows machine – both IE and Firefox will require some adjustments before they support SSO via Kerberos.

If this works for you, rejoice. If not, try setting the Apache log level to debug. Also, take SSL out of the equation by turning it off. At this point, you’ve now (hopefully) gotten single sign-on authentication working. Once everything is the way that you want it, you may want to add in a redirect that forces http:// traffic to https:// .

Now it’s time to work on group-based authorization. For that we need to configure mod_authnz_ldap.

Authorization with mod_authnz_ldap

Adding group support requires just a few changes. First, it is critical that you edit /etc/openldap/ldap.conf to add this line:

REFERRALS off

Do this now. Otherwise, this will never work and you will waste half your day trying to figure out why LDAP doesn’t work. Do not edit /etc/ldap.conf. On CentOS or RHEL, you must edit /etc/openldap/ldap.conf.

Next, add the following after the Kerberos settings in auth_kerb.conf, but still within the <Location> section:

AuthLDAPURL "ldaps://DOMAINSERVER.eu.emory.edu:636/dc=eu,dc=emory,dc=edu?cn?sub?(objectClass=user)" NONE
  AuthLDAPBindDN "cn=LDAPSERVICEACCOUNT,ou=users,ou=law,dc=eu,dc=emory,dc=edu"
  AuthLDAPBindPassword "LDAPSERVICEACCOUNTPASSWORD"
  require ldap-group cn=g law all-law,ou=global groups,ou=law,dc=eu,dc=emory,dc=edu

And that should do it.

Conclusion

Well, there you have it. Authentication via Kerberos and Authorization via groups in LDAP. This is harder to configure than one would expect. However, the result is very nice. Stay tuned for a short article on preparing Firefox and IE to take advantage of the Kerberized web server for SSO.

If this works for you, or even if it doesn’t, please let me know in the comments. Also, a  huge thank you to all of the people who were kind enough to document their work on this. Thanks!

Enhanced by Zemanta