Category: AIX

Tips and tricks for AIX

Posted on

Automatic creation of home directories

There are several possibilities under AIX to automatically create missing home directories when logging in. This is especially useful if the user accounts are managed through LDAP or another naming service and are not created locally. If a user is newly created in LDAP, he initially has no home directory on the AIX LDAP client:

$ ssh new_user@aix01
...
Could not chdir to home directory /home/new_user: No such file or directory
$ pwd
/
$ exit
$

Probably the easiest way to automatically create the home directory when logging in, is the attribute mkhomeatlogin in the file /etc/security/login.cfg. The default for this attribute is “false” if it is not set:

# lssec -f /etc/security/login.cfg -s usw -a mkhomeatlogin
usw mkhomeatlogin=
#

The attribute can be set to true with the chsec command:

# chsec -f /etc/security/login.cfg -s usw -a mkhomeatlogin=true
# lssec -f /etc/security/login.cfg -s usw -a mkhomeatlogin
usw mkhomeatlogin=true
#

We try the login again:

$ ssh new_user@aix01
...
$ pwd
/home/new_user
$

A new home directory has been created for the user.

Posted on

“who -r” does not return run level

On one of our systems, the command “who -r” did not return run level information. No error message was shown:

$ who -r
$ echo $?
0
$

As a consequence, an install script terminated with an error, since it was not able to determine the run level.

The information about the run level comes from the binary log file /etc/utmp. The run level is stored as the second record in this file. We assumed that /etc/utmp contained corrupt records.

The command /usr/sbin/acct/fwtmp (bos.acct) can be used to convert binary utmp-records to ASCII (and vice versa). The command expects the records to convert on standard input. In our case we got:

$ cat /etc/utmp | /usr/sbin/acct/fwtmp
                        system boot   2     0 0000 0000 1484666008                                  Tue Jan 17 16:13:28 CET 2017
root                                  0 804397248 0000 0000          0 \ufffd{\ufffd\ufffd                             Thu Jan  1 01:00:00 CET 1970
         naudio                       8 3473526 0000 0000 1484666008                                  Tue Jan 17 16:13:28 CET 2017
         naudio2                      8 3539068 0000 0000 1484666008                                  Tue Jan 17 16:13:28 CET 2017
...

The output above confirmed that the second record was corrupt, since it obviously did not contained the run level. Comparing with the entries from a working system showed how the correct records should look like:

                        system boot   2     0 0000 0000 1545044734                                  Mon Dec 17 12:05:34 2018
                        run-level 2   1     0 0062 0123 1545044734                                  Mon Dec 17 12:05:34 2018

First of all we made a copy of the corrupt /etc/utmp. Then we created an ASCII version using the above fwtmp command:

# cp /etc/utmp /etc/utmp.orig
# cat /etc/utmp | /usr/sbin/acct/fwtmp -X -L >/etc/utmp.ascii
#

The options -X and -L ensure that user and host names are not shortened.

Using an editor, we corrected the second entry by using the corresponding entry from the working system above. Then we corrected the timestamps by taking the values from the first entry. All in all the corrected version was:

                        system boot   2     0 0000 0000 1484666008                                  Tue Jan 17 16:13:28 CET 2017
                        run-level 2   1     0 0062 0123 1484666008                                  Tue Jan 17 16:13:28 CET 2017
         naudio                       8 3473526 0000 0000 1484666008                                  Tue Jan 17 16:13:28 CET 2017
...

Now we converted the corrected ASCII version back to the binary format and stored that version under /etc/utmp:

# cat /etc/utmp.ascii | /usr/sbin/acct/fwtmp -ic > /etc/utmp
#

Finally the command “who -r” worked again:

$ who -r
   .        run-level 2 Jan 17 16:13       2    0    S
$

The problem was resolved.

Posted on

/usr/sbin/rpm_share[440]: 36044986 Illegal instruction

The above error message showed up during the installation of an RPM package:

# rpm -U db4-4.7.25-2.aix5.1.ppc.rpm 
/usr/sbin/rpm_share[440]: 36044986 Illegal instruction
rpm_share: 0645-007 ATTENTION: get_rpm_inst_root_list() returned an unexpected result.
rpm_share: 0645-007 ATTENTION: update_inst_root() returned an unexpected result.

The rpm-command no longer works, a rebuild of the RPM database is therefore not possible anymore:

# rpm --rebuilddb
/usr/sbin/rpm_share[470]: 22478966 Illegal instruction

Reinstalling the fileset rpm.rte fixes the problem:

# installp -acFXYd . rpm.rte
+-----------------------------------------------------------------------------+
                    Pre-installation Verification...
+-----------------------------------------------------------------------------+

...

Installation Summary
--------------------
Name                        Level           Part        Event       Result
-------------------------------------------------------------------------------
rpm.rte                     4.13.0.3        USR         APPLY       SUCCESS    
rpm.rte                     4.13.0.3        ROOT        APPLY       SUCCESS

Afterwards the rpm-command works again:

# rpm -qa
...
db4-4.7.25-2.ppc
...
AIX-rpm-7.1.5.15-7.ppc