More on Sticky bits and ACLs

I'm going through the tedious task of hardening our Linux servers for an upcoming FSO visit as well as tweaking our existing kickstart for future builds.  A LOT of documentation goes into our kickstarts so we can pick things out and know what has been set and in response to which IAVA requirement.  I wanted a really clear example on how to properly use sticky bits and setfacl so to apply appropriate permissions when new files are created within a directory.  No point in setting the permissions on the current files if new ones can be created and bust the stig right?

The group ownership can be inherited by new files and folders created in your folder /path/to/parent by setting the setgid bit using chmod g+s like this:

chmod g+s /path/to/parent

Now, all new files and folder created under /path/to/parent will have the same group assigned as is set on /path/to/parent.

POSIX file permissions are not inherited; they are given by the creating process and combined with its current umask value.
However, you can use POSIX ACLs to achieve this. Set the default ACL on a directory:

setfacl -d -m u::rwX,g::rwX,o::- /path/to/parent

This will apply setfacl to the /path/to/parent directory, -modifying the -default ACLs – those that will be applied to newly created items. (Uppercase X means only directories will receive the +x bit.)

(If needed, you can add a u:someuser:rwX or g:someuser:rwX – preferably a group – to the ACLs.)

---

Note that, at least with ext3/ext4, you must mount the filesystem with the acl option, otherwise new ACLs cannot be set and existing ones will be ignored.

mount -o remount,acl /

Edit /etc/fstab to set this permanently.

Linux: sed command usage #1

I'm getting better with sed all the time but today was a challenging little booger.  I wanted to change /etc/pam.d/su from

cat /etc/pam.d/su
#%PAM-1.0
auth            sufficient      pam_rootok.so
# Uncomment the following line to implicitly trust users in the "wheel" group.
#auth           sufficient      pam_wheel.so trust use_uid
# Uncomment the following line to require a user to be in the "wheel" group.
#auth           required        pam_wheel.so use_uid
auth            include         system-auth
account         sufficient      pam_succeed_if.so uid = 0 use_uid quiet
account         include         system-auth
password        include         system-auth
session         include         system-auth
session         optional        pam_xauth.so

to

cat /etc/pam.d/su
#%PAM-1.0
auth            sufficient      pam_rootok.so
# Uncomment the following line to implicitly trust users in the "wheel" group.
#auth           sufficient      pam_wheel.so trust use_uid
# Uncomment the following line to require a user to be in the "wheel" group.
auth           required        pam_wheel.so use_uid
auth            include         system-auth
account         sufficient      pam_succeed_if.so uid = 0 use_uid quiet
account         include         system-auth
password        include         system-auth
session         include         system-auth
session         optional        pam_xauth.so

Notice the absense of the comment on line 6?  If you used sed -i 's/\#auth/auth/' <file> it would only look at the first line and then quit.  If you used sed -i 's/\#auth/auth/g' it'd change all of them.  I wanted to change just ONE out of the whole file.  The trick?  To specify a LINE for sed to work on.

sed -i '6s/\#auth/auth/' /etc/pam.d/su

Finally!  Looked for a while to find the answer.  The key is using the line number at the beginning of the search string here.  This link helped me get going in the right direction, http://www.linuxquestions.org/questions/linux-software-2/sed-display-text-on-specific-line-of-text-file-397405/. but it didn't give me EXACTLY what I wanted. However it helped me figure out how to specify a line.  I knew I could do it after I was able to print the line I wanted to work on.  SUCCESS!!

Need to make a Linux Kickstart iso?

I work with Linux on a daily basis.  I'm also fortunate to have a good grasp on how to build servers and most of my co-workers, aside from a few, have trusted me to make most of the servers.  It's a pain having to use a windows server just to mount the ISO to get the process started.  Also if you're working from a longer distance away the size of that ISO burdens your connection. Behold the necessary steps to make a tiny ISO of just what you need to get the process started!

http://mikent.wordpress.com/2012/04/12/how-to-create-a-kickstart-iso-boot-disk-for-redhat/

How to create a kickstart ISO boot disk for RedHat

1) logon as root
2) create a directory name bootdisk/RHEL
mkdir -p bootdisk/RHEL
3) copy the directory isolinux from your RedHat DVD or other location containing RedHat binaries in bootdisk/RHEL
example: cp -R /mnt/isolinux/* ~/bootdisk/RHEL/
4) change direcotry to ~/bootdisk/RHEL/
cd ~/bootdisk/RHEL/
5) create (or copy) your ks.cfg  (it will be discussed  later in another post how to create a kickstart file) in ~/bootdisk/RHEL/
example: cp ks.cfg ~/bootdisk/RHEL/
6) Now, you can create the ISO boot disk as follow (make sure you run the command from ~/bootdisk/RHEL/) :
mkisofs -r -T -J -V “RedHat KSBoot” -b isolinux.bin -c boot.cat -no-emul-boot -boot-load-size 4 -boot-info-table -v -o linuxboot.iso .
7) Burn your iso linuxboot.iso into a blank cd-rom or mount it as it is on a Virtual Machine for example
8) At linux boot prompt, type the following command:
linux ks=cdrom:/ks.cfg
if you need to install using a specific IP address using a specific Ks boot device, type the following:
linux ks=cdrom:/ks.cfg append ip=<IPaddress> netmask=<netmask> ksdevice=<NICx>
example: linux ks=cdrom:/ks.cfg append ip=10.10.10.10 netmask=255.255.255.0 ksdevice=eth0
9) your are done!

Disabling Adobe Updates on a Mac

As a Mac admin I'm becoming increasingly aware of all the crap you have to do special to administer these things in an enterprise environment.  It's not as easy as Active Directory is with GPOs but rather it's more linux like since it is a Unix OS based on BSD.  I have some annoying updaters which try to update all the friggin time Adobe products.  Finally found a site which tells me how to do it, Stop Adobe Update Manager Launching.

Now I just need to test it for deployment using either Workgroup Manager or ARD.

Disable Adobe Update Manager

You need to create a file called com.adobe.AdobeUpdater.Admin.plist in your ~/Library/Preferences/ which can be done in two different ways, through the Terminal with a defaults write command, or by manually creating the file with a text editor.
Here’s how to do it through the command line:
* Launch Terminal (located in /Applications/Utilities/)
* At the command prompt, paste this command exactly
defaults write com.adobe.AdobeUpdater.Admin Disable.Update -bool yes
* Hit return to execute the command and create the plist file
You can double-check that the file has been created by looking in ~/Library/Preferences/ for the file. Now, theoretically at least, Adobe Update Manager will not launch on Mac user login and system boot.
There is also the option of manually creating the plist file, either though a plist editor or if you’re familiar with plists just by using a text editor.
Again you need to create a file located at /Library/Preferences with the name com.adobe.AdobeUpdater.Admin.plist with a boolean set true to “Disable.Update”, as you can see in the screenshot below:

Now you should be able to reboot your Mac, logout of a user, and anything as usual, without the annoyance of Adobe Update Manager barging in on your computing session. Stopped in it’s tracks!

Hacking RPMs

Ever want to modify a file within an RPM but you don't have the spec file or possibly even the source rpm?  Now you can!  And I've done it successfully using the link included as a guide along with the man page.

http://magazine.redhat.com/2007/12/04/hacking-rpms-with-rpmrebuild/

It's ultra cool!  I've successfully built a customized RPM build for a zabbix agent which has modifications allowing it to be used across ANY network you'd like to use including a modified .conf file :)  Oh the joys of linux.

Hacking RPMs with rpmrebuild

by afsilva

A couple of months ago, I discovered a tool called rpmrebuild while searching for a way to reverse engineer the files installed on an older Fedora system back into its original RPM package. Rpmrebuild is able to reconstruct an RPM by looking up the information about it on the RPM database that is part of every RPM-based distribution like Fedora.


But rpmrebuild doesn’t stop there; you can also modify actual RPM packages without needing access to its SRPMS or even knowing much about SPEC files. Although this may not be recommended when dealing with core/base Linux system RPMS, it is incredibly useful for developers, release engineers, and system administrators who needs to create internal RPMs for their organizations.

For example, it is common practice for release engineers to have a “back out strategy” in case a release does not meet requirements during installation. With rpmrebuild, the version and release numbers of an RPM that may be replaced by a new one can be tweaked so that in case there is a failure and the ”back out” RPM is needed, the release engineers can simply install the back out RPMs over the new RPMs. Then the back out RPMs will have higher version and/or release numbers on them, so a tool like up2date or yum can automatically pick up on the changes.

Rpmrebuild is currently available for Fedora 7 and 8. To install:

yum install rpmrebuild

To rebuild an installed package in your system into an RPM:

rpmrebuild packagename

While rebuilding a package, rpmrebuild will let you know if files have been modified from their original state.

If they have, it will give you the option to continue or halt the rebuilding of the package, and it will ask you if you want to change the release number of the package.
Example:

[root@dhcp227-94 SOURCES]# rpmrebuild httpd
Processing files: httpd-2.2.3-7.el5
Wrote: /usr/src/redhat/RPMS/i386/httpd-2.2.3-7.el5.i386.rpm
result: /usr/src/redhat/RPMS/i386/httpd-2.2.3-7.el5.i386.rpm

My favorite feature of rpmrebuild is the ability to modify its spec file on the fly. By that I mean that you can actually edit the spec of an existing RPM without having to rebuild from source. Why is this useful? Well, you can modify RPM package requirements, change logs, descriptions, and other fields on the spec without having to go through the entire build process again. It can save you a lot of time if you are in the business of building RPMs and don’t necessarily use auto-builders like koji or buildbot.

Here how’s it done:

rpmrebuild -e -p –no-test-install package.rpm

• -e tells rpmrebuild you want to edit the whole spec file
• -p is used because we are editing an actual RPM file
• –notest-install stops rpmrebuild from auto-testing your RPM, just in case you are building an RPM on a workstation that does not have all required RPMs for that package

Rpmrebuild also offers certain shortcuts and plugins. Below I will change the release number of an RPM file without having to open up its spec file. This is a great for automating release numbering processes.

[root@dhcp227-94 i386]# rpmrebuild --release=99 -p --notest-install httpd-2.2.3-7.el5.i386.rpm
Processing files: httpd-2.2.3-99
Wrote: /usr/src/redhat/RPMS/i386/httpd-2.2.3-99.i386.rpm
result: /usr/src/redhat/RPMS/i386/httpd-2.2.3-99.i386.rpm

Notice that the httpd package went from release #7 to #99.

Not as recommended, but useful for your organization’s internal applications, you can modify the version number of an RPM as well:

rpmrebuild --change-spec-preamble='sed -e "s/^Version:.*/Version:1\.3\.1\.0\.1/"' --release=99 -p –notest-install  some-package-1.3.1-11.noarch.rpm

This command will rebuild your RPM and produce some-package-1.3.1.0.1-99.noarch.rpm.
Some other things to keep in mind about rpmrebuild are:
1.Once an RPM is rebuilt, it will lose its original signature (if signed).
2.You need to be root to rebuild a package only if there are root-protected files in that package.
3.Rpmrebuild will respect your RPM “home” building location, so if you have .rpmmacros set up in your home dir, your rebuilt RPMs will show up there.

The authors of rpmrebuild, Eric Gerbier and Valery Reznic, point out that even though the newer versions of RPM have a repackage option, they still require the user to uninstall that package from their system, which sometimes is not necessarily easy because of the dependencies on that package.
If you want to rebuild an old RPM that is not easily available on the Internet anymore, or if you need to tweak packages for your organization’s internal releases, or even if all you want to do is study and learn a bit more about RPM packaging, rpmrebuild is a great tool to have.

D-Link Business Switches and port trunking

Aggravating as all get out when a business product which is CAPABLE of doing normal network switching decides to do things differently.  I can understand maybe Cisco doing this or Juniper because they have the clout to change things.  But D-Link?  Really guys, why must you try to change how port trunking or vlan tagging works?

I'm configuring the network at my school with vlans, and it works great on the wireless, but wired connections... SHEESH.  Short of it all is I did it wrong the first time.  Had to go find someone else asking the same question on D-Link's support documents to find the correct answer.

With these switches you have Untagged, Tagged, and Not a Member settings for each port.
"Not member": This port is not a member of the VLAN.
"Tagged": The packets have already a VLAN-tag, i.e. they are tagged by the network device connected to this port.
"Untagged": The packets at this port have no VLAN-tags, so the incoming packets are tagged by the switch and the outgoing packets are untagged by the switch.

Which means you have to instead select the port and REMOVE all the OTHER vlans from the port instead of just add the vlan to the port you wish it to be a member of.  That's bass-ackwards.  So for the rest of the day my network is not working the way it was designed to work and I have to go AFTER work to fix it. GRR!!!

For more context see here: VLANS - Tagged, untagged, what do they all mean?

Annoying Internet Explorer 8 Start Menu Icon

Ever installed Internet Explorer 8 on a Windows XP machine and if you didn't ONLY install that program before you rebooted and you kept using Internet Explorer you hit the start menu and the icon is missing?!  Pain in the rear if you don't know how to find the answer.  Luckily this has happened to me for the last time. Below is the solution.

Try this:

Open the registry editor, navigate you way to HKEY_LOCAL_MACHINE\SOFTWARE\Clients\ StartMenuInternet\IEXPLORE.EXE\DefaultI con

It should look something like C:\Program Files\Internet Explorer\iexplore.exe,-7

All you have to do is to change the value to
C:\Program Files\Internet Explorer\iexplore.exe,-32528

Log-off or restart and Viola! This will change the existing IE icon to the current IE 8 icon (with the yellow swish). If you like the "classic" IE icon, change the value to 0.

The beginning.

I'm expecting this to be a lessons learned blog for the most part. I am a Linux Admin, Pilot, Network Administrator, cyclist, and many others.  I plan to update this whenever I get an "ah HA" moment.  I really do learn a lot each day and some lessons I don't always remember the finer details.  Luckily Google is there to retrieve the information for me the second and some times third time but I'd like to try to make a secondary memory of sorts for those types of things.  A digital braindump if you will.

Today's lesson: cp -p will copy a file while retaining prior permissions.

(hey i didn't say these things were earth shattering or exciting.)