Computational groups for Leopard directory services

There are occasionally times in system administration where I feel as if I’m on the bleeding edge. They are not frequent … which probably puts me somewhere much closer to the lunatic fringe. These are also the times when no one else probably cares about the same things — but no matter, I generally have my own pragmatic reasons for them. This is one of those times, and I have waited for this particular feature for a while. I'm happy to have discovered it.

Let’s set up a scenario: you have a Mac OS X Tiger system bound to an Active Directory. All of the thousands of Active Directory users are available on that Tiger system; any one of them could log in to that computer. This isn’t bad if you only allow logins at the login window. However, if you enable any sharing services, the game changes and your exposure widens.

What if you don’t want everyone in your enterprise directory to log in? This gets a little scary to think about. What if you want to limit those who can log in, particularly for services like SSH? How do you let some users in but keep others out — even going to the point of denying access to whole classes of users?

There are a number of things you can do, some at the client and some in the directory. Unfortunately, some kinds of configuration must be done at the client. There isn’t a great way I’m aware of to distinguish between local and network accounts, complicates the situation for the administrator. With Open Directory in Tiger, any user from any directory service was as valid as any other. There will probably be some manual effort — ongoing effort at that — required to restrict access.

Enter Mac OS X Leopard. It now has “computational groups” in its Open Directory client. They are like smart groups in your local node. The memberships of these groups are calculated, rather than static. They are more like query-based groups.

For the scenario above, the two computational groups that provide the most utility are probably “localaccounts” and “netaccounts.” They allow you to distinguish users who are in the DSLocal database from those in Active Directory. This magic gives the glue we need to maintain finer-grained access controls automatically over time; you could use these groups for the sshd server configuration or in the authorization database at /private/etc/authorization, as examples.

Let’s try the groups out with dsmemberutil, with which we can determine whether or not a user is a member of a specific group. What about the first local user, typically created by the Mac OS X Setup Assistant?

$ dsmemberutil checkmembership -u 501 -G localaccounts
user is a member of the group
$ dsmemberutil checkmembership -u 501 -G netaccounts
user is not a member of the group

Digging further, all of the computational groups I’ve found had the same comment. We can search for that to find the rest, to find all of the computational groups that ship with a Leopard client.

$ dscl . -search /Groups Comment "Group membership calculated by system" | awk '/user|account/ { print $1 }'

From that list, we can determine more information about the groups:

Short name Real name GID Notes
authedusers Authenticated Users 50
consoleusers Terminal Server User 53
interactusers Interactive 51
netaccounts Network Accounts 61 Accounts from a non-local database
localaccounts Local Accounts 62 Accounts residing only in one of the local databases
netusers Network 52

Unfortunately, I couldn’t find any other documentation about the computational groups. If you do find some, I’d love to hear about it.

Get the UUID for a user quickly with dsmemberutil on Leopard

Never let it be said that Apple’s Open Directory command lines tools make complete sense. After all, why would you think to use the new dsmemberutil tool in Leopard to find the Universally Unique Identifier (UUID) for a user account?

You would have if you had read the first line of its man page, maybe, where it outlines exactly how miscellaneous — and I say that in the kindest possible way — the command is: “dsmemberutil -- various operations for the membership APIs.”

Enough lead in; the UUID for a user (as well as a group) can be found with this tool:

$ dsmemberutil getuuid -u 501

Choose a user with either -u with the userid, or -U with the short username.

One way this can be exceedingly useful is if you want to track down the shadow password file corresponding to a specific user account:

$ sudo ls -1 /private/var/db/shadow/hash | grep dsmemberutil getuuid -u <em>501</em>

Hey, look, the UUIDs match, saving a trip to dscl!

Dissecting group membership with dsmemberutil in Leopard

There are times when you need to determine whether a user is a member of a group or not. Knowing the membership of a group can have an impacts on security or operation of system administration scripts. The dsmemberutil tool in Mac OS X Leopard lets you perform checks like this on the command line or in scripts.

You feed it a group and a user, and the tool tells you whether or not the user is a member of the group. This works with group nesting, thanks to the wonder of the memberd functionality first available in Mac OS X Tiger.

$ dsmemberutil checkmembership -u 501 -G admin
user is a member of the group
$ dsmemberutil checkmembership -u 502 -G admin
user is not a member of the group

Unfortunately, you must parse the output, which hinges on whether “not” appears in the text. The exit code for both commands above is “0,” for success.

Compiling Rsync 3 as a Universal Binary for Leopard

When I come across software I might need to add into Mac OS X that requires compilation, I typically want to produce one Universal Binary. Make it a four-way UB and you get both 32- and 64-bit support.

A single binary is ideal for a Radmind transcript (or other package, if you wanted to bundle it into an installer) that can be deployed on both PowerPC and Intel Macs on Leopard.

Since rsync 3.0.2 with some patches is apparently working quite well at preserving Mac OS X data and metadata — passing the Backup Bouncer tests — I thought I'd try my hand at a four-way Universal Binary.

What worked for me, using a Mac Pro 4x2.8 GHz with Mac OS X 10.5.2 and Xcode 3.0, was to start with Mike Bombich's instructions and modify them with some fairly standard Universal Binary build steps. The configure and compile were both less than a minute on this system.

$ ./prepare-source
$ env CFLAGS="-O -g -isysroot /Developer/SDKs/MacOSX10.5.sdk -arch i386 -arch ppc -arch ppc64 -arch x86_64" \
LDFLAGS="-arch i386 -arch ppc -arch ppc64 -arch x86_64" \
./configure --prefix=/usr/local --disable-dependency-tracking

$ make
$ sudo make install

I have seen the use of "-Wl,-syslibroot,/Developer/SDKs/MacOSX10.5.sdk," in the LDFLAGS environment variable when compiling some applications but this did not work for me with rsync; when I removed it, rsync 3.0.2 configured successfully for me.

The result of the above build process appears to be a full four-way UB:

$ lipo -info /usr/local/bin/rsync
Architectures in the fat file: /usr/local/bin/rsync are: i386 ppc7400 ppc64 x86_64

A local transfer on the build system appears to have worked correctly. I did not test with Backup Bouncer, sync with a non-Mac system, or when shuttling data between architectures. So, accept these results with a grain of salt; I’m just happy I got rsync to compile for now.

Manipulate Apple Type Services settings in Leopard with atsutil

Aaron Hall noted+on+the+MacEnterprise+list that the font caches in Leopard “seem to be under /var/folders now, but you can manipulate them with the cool new atsutil in Leopard.”

The atsutil command is quite an interesting find to me, so I wanted to point it out.

Microsoft Office 2008 Service Pack 1 released

Microsoft released Service Pack 1 to update Office 2008 for Macintosh on Tuesday, May 13. The update makes a significant number of changes — spread across Entourage, Word, Excel, and PowerPoint — so if you are a system administrator, I would recommend examining the full description of Office 2008 for Mac Service Pack 1 (12.1.0) in its release notes.

The update is downloadable via Microsoft AutoUpdate (invoke that helper program with Help > Check for Updates in any of the Office applications, or look inside /Library/Application Support/Microsoft in the file system) as well as the Microsoft Office for Mac Web site.

List directory nodes available to a local Mac OS X host with dscl

It took me a little while to figure out how I could tell what DirectoryService directory nodes are available to a Mac OS X system. I could perform the task interactively, just issuing an “ls” command at the dscl “>” prompt. But for a non-interactive one-liner, I was bedeviled by what I thought was a pretty basic function for the dscl tool.

I wanted to find all of the top-level nodes — because, for example, that could tell+me+if+I+really+had+access+to+an+Active+Directory+or+not.

It turns out that I was missing the use of “localhost” as the datastore parameter, which reliably returns the information I wanted in a parsable form. Rather than “/” (simply doesn’t work) or just a period (only uses the local directory) — or even listing "/Active\ Directory" explicitly — “localhost” specifies all of the directories available to the local host. I was incorrectly assuming “localhost” would tie me to the local directory, as with “localonly,” and I thus avoided trying it until I felt like an utter failure.

$ dscl localhost -list .

This produces the following output on a Leopard-based system which only has local accounts in DSLocal:

$ dscl localhost -list .

If your system happens to be bound to a Microsoft Active Directory, you’ll instead see it prepended:

$ dscl localhost -list .
Active Directory

Or if you’re on Tiger — again bound to Active Directory — you’ll still see “Bonjour,” “NetInfo,” and “SLP” in the mix:

$ dscl localhost -list .
Active Directory

Example of mismatched UIDs between a calendar event and cancellation in Microsoft Entourage

I had a good example today of how mismatched UIDs between Exchange meeting updates and existing events can affect Microsoft Entourage’s processing of those incoming updates.

I used the same drag-and-drop-to-export methodology as in my previous article. The iCalendar .ics file for the existing event that I dropped out of my Entourage calendar contained these unique identifiers:


The e-mail cancellation from the meeting organizer, when dropped into the Finder as a .eml file, had the following unique identifiers:


As we can see, even though these were supposedly for the same event, neither the UID nor the X-ENTOURAGE_UUID properties matched. I emphasize the UID property because, as per my post on matching UIDs between Exchange meeting updates and existing events in Microsoft Entourage, I believe it to be the most important.

Both I and the meeting organizer are Entourage users in this case. That may be why both the event and the cancellation had X-ENTOURAGE_UUID properties at all. The UID property has appeared consistently for me in both events and their updates, no matter what their source.

Because of the discrepancy between unique identifiers, Entourage apparently couldn’t determine whether the cancellation was for the event that existed in my calendar. When Entourage “processed” the cancellation, it removed nothing from my calendar.

This seems entirely reasonable, although I would personally like Entourage to do deeper matching if it could: how about “is there an existing event by the same organizer and/or with the same name at the time of the update or cancellation?” Matching UIDs has got to be simpler.

How does this happen? Well, for me, it could have happened any number of ways: There could have been a past error when:

  • Entourage was trying to synchronize to my Exchange account
  • my Entourage calendar was synchronizing with Sync Services on Tiger or Leopard
  • Missing Sync for Palm OS, which I use with my Palm Treo 650, was performing its synchronization with Sync Services on Tiger or Leopard
  • I was using Microsoft Outlook in cached mode with my Exchange account
  • I used Duplicate Killer for Outlook from 4Team to remove massive numbers of duplicate calendar entries from my Outlook/Exchange calendar (which could have resulted in the loss of events with their original UIDs).

As far as I can determine, when you have data synchronization problems, you’re going to have one of the following outcomes:

  • data is added
  • data is deleted
  • data is modified
  • nothing happens.

Not all of these are pleasant, mind you — and that can depend on the circumstances involved. Deleting data is perfectly acceptable in some scenarios where it needed to be removed, but not others where it means it has just been lost. Adding data is great, unless it results in unwanted duplicates.

Anyway, I hope this helps you understand how Microsoft Entourage works a bit better.

Flat panel wake up and on screen display delay

Now that I have my dual-monitor Gateway FPD2485W setup, I’ve got a few complaints. Of course!

The two monitors take an awfully long time to wake up from their power-saving mode. Then, when they finally wake up — invariably at different times, the newer one first — I get the on-screen display (OSD) overlay telling me that they’ve chosen to accept the DVI input.

Well, duh, that’s the only video source hooked up to them, so it’s not helpful. This wouldn’t be so bad, but the OSD overlay stays on the screen for what seems like eons. Since the overlays are smack dab in the middle of the screen and are opaque, they block important visual elements like Mac OS X’s login window.

So far, I haven’t been able to find out how to get rid of the OSD overlay. If I could do that, I think I’d tolerate the wake up delay more readily.

Certainly, some of this is Gateway’s fault. I guess I can’t blame them much since they specifically don’t support Macs and that’s what I’ve hooked the flat panels up to. I could have gone with some brand that did advertise Mac support, but I didn’t. This must be my payback. Grin.

Mac OS X keychain and password storage

I’ve found that trying to explain the Mac OS X keychain at all tends to make peoples’ eyes glaze over. The keychain is poorly-understood overall, perhaps because it tries to bridge the gap between security and convenience.

A few thoughts:

  • A keychain has its own password which may or may not be set to the same as the password for the login account.
  • The keychain password is completely independent of the login account’s password, even if it is the same text as the login account’s password. They can be changed independently. When they are the same, they are just two passwords which happen to be the same.
  • User keychains are created within user home directories, and are protected by file system permissions while they are enforced.
  • Keychain keys are further protected by 3DES encryption. Directory or metadata information is in cleartext.
  • The Mac OS X Setup Assistant and Accounts System Preferences both create accounts whose login passwords and keychain passwords will match.
  • If the password is shared between the login account and that account’s default keychain, the keychain will be unlocked during the login process. This is the default for accounts created by the Mac OS X Setup Assistant and Accounts System Preferences.
  • If the default keychain’s password does not match the login account’s password, the keychain will not be unlocked automatically during the login process. The user may be prompted to unlock it, using the keychain password, if other applications require a key stored within.
  • The only time that a password change for a login account changes that user’s default keychain password is when the login account is logged in and changes its own password through Accounts System Preferences.
  • If the computer is bound to a directory service, a login account may be tied to that. However, the keychain is not. Changing the login account’s password through a directory service does not reset the keychain’s password. The keychain’s existing password will remain until or unless it is changed.
  • A third-party software utility, Keychain Minder, can help to keep login and keychain passwords in sync, if desired. This may be especially helpful in a directory service environment, where you are more likely to change account passwords externally rather through Mac OS X’s built-in means. It also provides an opt-out capability for those who specifically want different login and keychain passwords.
  • If the computer is bound to a directory service, and a directory service-based login account was compromised, there is a chance that the password for the default keychain in that account is also compromised. Changing the password for the login account in the directory service will protect the login account. However, that will not necessarily protect the keychain stored within the account’s home directory on disk. Whether or not the keychain password was the same as the former password for the login account — the keychain’s password should probably also be changed.
  • The long-term use of the same password for a keychain can be a risk; as it gets stale, it lessens the protection on each key in the keychain.
  • There is currently no policy enforcement mechanism, akin to pwpolicy, for keychain passwords.
Syndicate content