Fix bug in exception handling where we were sending the wrong thing as detail.
Basically we were catching an LDAP error, generating an IPAError from it,
catching that, then setting the detail of the 2nd exception to another IPAError
rather than the root exception. This caused anything looking at e.detail to
crap out
Resolves 432136
No longer create a PKCS#12 file that contains the CA
No longer send the entire CA to each replica, generate the SSL certs on master
Fix number of bugs in ipa-replica-install and prepare
Produce status output during replica creation
If you run ipa_generate_password() multiple times, one
after the other, then you get the same password each time.
This is because it uses the current time to seed the
pseudo random number generator.
The easiest solution is to just use the default method
which seeds itself from /dev/urandom if available,
and uses a fractional time value otherwise.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
The ipa_webgui and ipa_kpasswd instance code is identical
and I want to add another similar instance down the line,
so re-factor the code into a service.SimpleServiceInstance
class.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
DsInstance.pkcs12_info isn't currently initialised in
the constructore so, e.g. __enable_ssl() assumes that
create_instance() has initialised it.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
If, in future, we change the server ID so that it's not
derived from the realm name, there's a fair few places
that need to be changed.
Make that easier by having config_dirname() take the
server ID rather than the realm name. That makes sense
anyway so we don't have to realm_to_serverid() so
much.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Let's assume that all ipaserver.dsinstance could be used
somewhere where asking questions on stdout/stdin is not
approriate and re-factor the code to be suitable in
those situations too.
i.e. make check_existing_installation() return a list of
server IDs and make check_ports() return an (unsecure,
secure) tuple indication which ports are in use.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
