All the headers are now listed in _SOURCES variable.
It seems weird but this is what GNU Automake manual suggests in section
9.2 Header files:
Headers used by programs or convenience libraries are not installed.
The noinst_HEADERS variable can be used for such headers.
However when the header actually belongs to a single convenience library
or program, we recommend listing it in the program’s or library’s
_SOURCES variable (see Program Sources) instead of in noinst_HEADERS.
This is clearer for the Makefile.am reader.
noinst_HEADERS would be the right variable to use in a directory containing
only headers and no associated library or program.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Static files from Git which are not touched by the build system
have to be explicitly listed in Makefile.am so they get into tarball.
EXTRA_DIST lists whole sub-directories with static files.
This is not ideal but we do not have enough time to fix it properly.
Dojo builder patch files were renames to shorten their name.
The original names were exceeding autotools length limit.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Static files from Git which are not touched by the build system
have to be explicitly listed in Makefile.am so they get into tarball.
ipa script was missing on installed systems for the same reason.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
This is step forward working VPATH builds which cleanly separate sources
and build artifacts. It makes the system cleaner and easier to
understand.
Python and web UI likely require more work to make VPATH builds working.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Most of the logic was unnecessary and wrong. This caused make install to fail.
This commit removes unnecessary declarations and creates static library
which is not installed. make install in asn1 subdirectory is now passing
(and doing nothing).
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
This version builds only one version of Python packages. If you want to
build for Python 2 & 3 call configure twice using different --with-python
or specify PYTHON variable when calling make.
dist-hook is using SOURCES.txt file from egg-info.
According to Petr Viktorin this should be enough for our purposes
and avoids need to create plugins for setuptools.
Currently VPATH builds do not work for various reasons.
This should be fixed later on.
Most credit goes to these guys:
Christian Heimes <cheimes@redhat.com>
Petr Viktorin <pviktori@redhat.com>
Kevin Brown <kevin@kevin-brown.com>
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Version information is now in VERSION.m4 instead of VERSION.
Makefile target version-update was minimized and configure can be run
before make. Makefile temporarily contains hardcoded version which has
to match the one specified in VERSION.m4.
This is preparatory step which will allow us to replace hand-made
Makefile with one generated by Automake.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
The original approach with __path__ implemented
by 8f98fa1bd5 broke Pylint:
We decided to resort back to symlinks as it is easiest solution
which does not break pylint in weird ways.
This commit introduces configure --with-ipaplatform option.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Lukas Slebodnik <lslebodn@redhat.com>
Reviewed-By: Stanislav Laznicka <slaznick@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
This is temporary workaround necessary until we throw away the hand-made
Makefile. ipaplatform is going to be managed by configure, but configure
right now depends on version-update target in Makefile.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Lukas Slebodnik <lslebodn@redhat.com>
Reviewed-By: Stanislav Laznicka <slaznick@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
This is temporary workaround. makeapi/makeaci transitively import
ipaplatform, which will not exist before configure is executed.
On the other hand, configure requires version.m4,
which is generated by current Makefile.
This change works around this chicked-egg problem. It will disappear
when we start using top-level configure to generate top-level Makefile.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Lukas Slebodnik <lslebodn@redhat.com>
Reviewed-By: Stanislav Laznicka <slaznick@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Use package config instead of checking headers.
Package config is faster because it does not invoke compiler
and guarantees proper linking flags because these are provided
by package maintainer instead of hardcoded into build system.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Lukas Slebodnik <lslebodn@redhat.com>
Reviewed-By: Stanislav Laznicka <slaznick@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Use package config instead of checking headers.
Package config is faster because it does not invoke compiler
and guarantees proper linking flags because these are provided
by package maintainer instead of hardcoded into build system.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Lukas Slebodnik <lslebodn@redhat.com>
Reviewed-By: Stanislav Laznicka <slaznick@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Use package config instead of checking headers.
Package config is faster because it does not invoke compiler
and guarantees proper linking flags because these are provided
by package maintainer instead of hardcoded into build system.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Lukas Slebodnik <lslebodn@redhat.com>
Reviewed-By: Stanislav Laznicka <slaznick@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Use package config instead of checking headers.
Package config is faster because it does not invoke compiler
and guarantees proper linking flags because these are provided
by package maintainer instead of hardcoded into build system.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Lukas Slebodnik <lslebodn@redhat.com>
Reviewed-By: Stanislav Laznicka <slaznick@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Top-level Makefile is still not managed by Automake (e.g. hand-made).
This is preparatory work. Other configure.ac files will be gradually
merged into the top-level one. After that we will be able to throw-away
the hand-made top-level Makefile and use Automake for everything.
https://fedorahosted.org/freeipa/ticket/6418
Reviewed-By: Lukas Slebodnik <lslebodn@redhat.com>
Reviewed-By: Stanislav Laznicka <slaznick@redhat.com>
Reviewed-By: Christian Heimes <cheimes@redhat.com>
Interestingly, the new macro AC_HEADER_STDC is alredy present.
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Alexander Bokovoy <abokovoy@redhat.com>
The Python detection logic will be improved later when we start to use
top-level configure.ac to manage build completely. For now simple bump
is enough.
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Alexander Bokovoy <abokovoy@redhat.com>
Mozldap is not used for some time now. We can remove
all traces of it.
AFAIK the complex logic for OpenLDAP detection should not be
necessary and -lldap_r -llber options should suffice.
Unfortunatelly OpenLDAP package does not ship
package config files so we have to hardcode flags.
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Alexander Bokovoy <abokovoy@redhat.com>
Use package config instead of checking headers.
Package config is faster because it does not invoke compiler
and guarantees proper linking flags because these are provided
by package maintainer instead of hardcoded into build system.
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Alexander Bokovoy <abokovoy@redhat.com>
Use package config instead of checking headers.
Package config is faster because it does not invoke compiler
and guarantees proper linking flags because these are provided
by package maintainer instead of hardcoded into build system.
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Alexander Bokovoy <abokovoy@redhat.com>
Use package config instead of checking headers.
Package config is faster because it does not invoke compiler
and guarantees proper linking flags because these are provided
by package maintainer instead of hardcoded into build system.
libkrad does not have package config file so we keep the old way here.
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Alexander Bokovoy <abokovoy@redhat.com>
It was working accidentally because krb5 libs are part of OPENLDAP_LIBS.
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Alexander Bokovoy <abokovoy@redhat.com>
This is to prevent situations when installer prints
CRITICAL Failed to load ....ldif
and continues just to crash later on because of non-existing LDAP container
or so on.
Reviewed-By: Martin Basti <mbasti@redhat.com>
Errors during DNS resolution might indicate that forwarder is the
necessary configuration which is missing. Now we disallow adding a
forwarder only if the zone is normally resolvable without the forwarder.
https://fedorahosted.org/freeipa/ticket/6062
Reviewed-By: Martin Basti <mbasti@redhat.com>
Previously, update_dnsforward_emptyzones failed with an exeception if
DNS query failed for some reason. Now the error is logged and upgrade
continues.
I assume that this is okay because the DNS query is used as heuristics
of last resort in the upgrade logic and failure to do so should not have
catastrophics consequences: In the worst case, the admin needs to
manually change forwarding policy from 'first' to 'only'.
In the end I have decided not to auto-start BIND because BIND depends on
GSSAPI for authentication, which in turn depends on KDC ... Alternative
like reconfiguring BIND to use LDAPI+EXTERNAL and reconfiguring DS to
accept LDAP external bind from named user are too complicated.
https://fedorahosted.org/freeipa/ticket/6205
Reviewed-By: Martin Basti <mbasti@redhat.com>
ipa-replica-prepare must be run on a replica with CA or all the certs
needs to be provided (for CA-less case).
The old messages were utterly confusing because they mixed errors about
missing certs and missing local CA instance into one text.
https://fedorahosted.org/freeipa/ticket/6134
Reviewed-By: Martin Basti <mbasti@redhat.com>
Reviewed-By: Ben Lipton <blipton@redhat.com>
Previously resolver was returning CheckedIPAddress objects. This
internal server error in cases where DNS actually returned reserved IP
addresses.
Now the resolver is returning UnsafeIPAddress objects which do syntactic
checks but do not filter IP addresses.
From now on we can decide if some IP address should be accepted as-is or
if it needs to be contrained to some subset of IP addresses using
CheckedIPAddress class.
This regression was caused by changes for
https://fedorahosted.org/freeipa/ticket/5710
Reviewed-By: Martin Basti <mbasti@redhat.com>
All these were unused so I'm removing them to keep the code clean and
easier to read. At this point it is clear that only difference between
netaddr.IPAddress and CheckedIPAddress is prefixlen attribute.
Reviewed-By: Martin Basti <mbasti@redhat.com>
Previously the installer did not reinitialize resolver so queries for
records created using --ip-address option might not be answered. This led
to incorrect results during 'Updating DNS system records' phase at the
end of installation.
This is kind of hack but right now we do not have enough time to extend
python-dns's interface with resolver_reinit() method.
https://fedorahosted.org/freeipa/ticket/5962
Reviewed-By: Martin Basti <mbasti@redhat.com>
Previously we were checking content of DNS before actually adding DNS
records for replicas. This is causing cycle in logic and adds weird
corner cases to the installer which can blow up on DNS timeout or so.
The check was completely unnecessary because the installer knows IP
addresses and name of the machine. Removal of the check makes
the installer more reliable.
https://fedorahosted.org/freeipa/ticket/5962
Reviewed-By: Martin Basti <mbasti@redhat.com>
Realmdomains integration into DNS commands pre-dates split of DNS forward zones
and DNS master zones into two distinct commands.
There was an forgotten condition in dnszone_add command which caused omission
of DNS master zones with non-empty forwarders from realmdomain list.
https://fedorahosted.org/freeipa/ticket/5980
Reviewed-By: Martin Basti <mbasti@redhat.com>
__add_ca_records_from_hostname() now skips over DNS exceptions and
retries resolution until timeout of 120 seconds is reached.
Luckily current logic fails safe: In cases where resolution failed for
all the CA servers, the resulting zone object will not contain ipa-ca
record at all and the update logic will skip update for this name.
I.e. the original values in ipa-ca record set will be left in place.
https://fedorahosted.org/freeipa/ticket/2008
Reviewed-By: Martin Basti <mbasti@redhat.com>
This change is necessary to override automatic empty zone configuration
in latest BIND and bind-dyndb-ldap 9.0+.
This upgrade has to be done on each IPA DNS server independently.
https://fedorahosted.org/freeipa/ticket/5710
Reviewed-By: Martin Basti <mbasti@redhat.com>
This change is necessary to override automatic empty zone configuration
in latest BIND and bind-dyndb-ldap 9.0+.
This procedure is still not complete because we need to handle global
forwarders in named.conf too (independently on each server).
https://fedorahosted.org/freeipa/ticket/5710
Reviewed-By: Martin Basti <mbasti@redhat.com>
This change is necessary to override automatic empty zone configuration
in latest BIND and bind-dyndb-ldap 9.0+.
This procedure is still not complete because we need to handle global
forwarders too (in LDAP and in named.conf on each server).
https://fedorahosted.org/freeipa/ticket/5710
Reviewed-By: Martin Basti <mbasti@redhat.com>
Ad-hoc LDAP calls in DNS upgrade code were hard to maintain and
ipaConfigString was bad idea from the very beginning as it was hard to
manipulate the number in it.
To avoid problems in future we are introducing new ipaDNSVersion
attribute which is used on cn=dns instead of ipaConfigString.
Original value of ipaConfigString is kept in the tree for now
so older upgraders see it and do not execute the upgrade procedure again.
The attribute can be changed only by installer/upgrade so it is not
exposed in dnsconfig_mod API.
Command dnsconfig_show displays it only if --all option was used.
https://fedorahosted.org/freeipa/ticket/5710
Reviewed-By: Martin Basti <mbasti@redhat.com>
The code was duplicate and less generic anyway.
As a side-effect I had to re-wrap dns.exception.DNSException into a
PublicError so it can be displayed to the user.
DNSError is now a super class for other DNS-related errors. Errors from
DNS resolver are re-raised as DNSResolverError.
https://fedorahosted.org/freeipa/ticket/5710
Reviewed-By: Martin Basti <mbasti@redhat.com>
After discussion with Martin Basti we decided to standardize on root_logger
with hope that one day we will use root_logger.getLogger('module')
to make logging prettier and tunable per module.
https://fedorahosted.org/freeipa/ticket/5710
Reviewed-By: Martin Basti <mbasti@redhat.com>
Forwarding policy "first" or "none" may conflicts with some automatic empty
zones. Queries for zones specified by RFC 6303 will ignore
forwarding and recursion and always result in NXDOMAIN answers.
This is not detected and warned about. Global forwarding is equivalent
to forward zone ".".
Example:
Forward zone 1.10.in-addr.arpa with policy "first"
will not forward anything because BIND will automatically prefer
automatic empty zone "10.in-addr.arpa." which is authoritative.
https://fedorahosted.org/freeipa/ticket/5710
Reviewed-By: Martin Basti <mbasti@redhat.com>
Petr Spacek