Files
freeipa/install/updates
Alexander Bokovoy 3692a1c57f trusts: harden trust-fetch-domains oddjobd-based script
When ipa-getkeytab is used to fetch trusted domain object credentials,
the fetched entry has always kvno 1. ipa-getkeytab always adds a key to
keytab which means older key versions will be in the SSSD keytab and
will confuse libkrb5 ccache initialization code as all kvno values are
equal to 1. Wrong key is picked up then and kinit fails.

To solve this problem, always remove existing
/var/lib/sss/keytabs/forest.keytab before retrieving a new one.

To make sure script's input cannot be used to define what should be
removed (by passing a relative path), make sure we retrieve trusted
forest name from LDAP. If it is not possible to retrieve, the script
will issue an exception and quit. If abrtd is running, this will be
recorded as a 'crash' and an attempt to use script by malicious user
would be recorded as well in the abrtd journal.

Additionally, as com.redhat.idm.trust-fetch-domains will create
ID ranges for the domains of the trusted forest if they don't exist,
it needs permissions to do so. The permission should be granted only
to cifs/ipa.master@IPA.REALM services which means they must have
krbprincipalname=cifs/*@IPA.REALM,cn=services,... DN and be members of
cn=adtrust agents,cn=sysaccounts,... group.

Solves https://bugzilla.redhat.com/show_bug.cgi?id=1250190

Ticket https://fedorahosted.org/freeipa/ticket/5182

Reviewed-By: Tomas Babej <tbabej@redhat.com>
2015-08-18 18:48:12 +02:00
..
2015-06-11 10:50:31 +00:00
2015-06-04 08:27:33 +00:00
2015-06-11 10:50:31 +00:00
2015-06-11 10:50:31 +00:00

The update files are sorted before being processed because there are
cases where order matters (such as getting schema added first, creating
parent entries, etc).

Updates are applied in blocks of ten so that any entries that are dependant
on another can be added successfully without having to rely on the length
of the DN to get the sorting correct.

The file names should use the format #-<description>.update where # conforms
to this:

10 - 19: Configuration
20 - 29: 389-ds configuration, new indices
30 - 39: Structual elements of the DIT
40 - 49: Pre-loaded data
50 - 59: Cleanup existing data
60 - 69: AD Trust
70 - 79: Reserved
80 - 89: Reserved

These numbers aren't absolute, there may be reasons to put an update
into one place or another, but by adhereing to the scheme it will be
easier to find existing updates and know where to put new ones.