BIND 10 #1643: TSIG configuration syntax should be as consistent as possible for auth and xfrout

[BIND 10 Development]

#1643: TSIG configuration syntax should be as consistent as possible for auth and
xfrout
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:  jinmei
  jinmei                             |                Status:  reviewing
                       Type:         |             Milestone:
  defect                             |  Sprint-20120306
                   Priority:  major  |            Resolution:
                  Component:         |             Sensitive:  0
  configuration                      |           Sub-Project:  DNS
                   Keywords:         |  Estimated Difficulty:  5
            Defect Severity:  N/A    |           Total Hours:  0
Feature Depending on Ticket:         |
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------
Changes (by vorner):

 * owner:  vorner => jinmei


Comment:

 Hello

 Replying to [comment:7 jinmei]:
 > Ideally, I'd like to confirm actual zone transfer using TSIG via
 > system tests.  Since configuration involves multiple processes, and
 > also we now have additional indirection to the global key ring,
 > I'm afraid it's more likely to have a system-level bug that cannot be
 > detected via unittests.  This could be a separate deferred ticket
 > though.

 OK, I'll add the ticket when merging this.

 > - Please explain a bit more rationale about this change.  It's not
 >   crystal clear.  (maybe we need to add the implication to the
 >   add_remote_config() description)
 > {{{
 >     * The config callback should be called after the module is ready.
 > }}}

 The _add_remote_config_internal did call the callback if there were non-
 default values. However, it did before setting the internal structures, so
 it complain the remote config is not set up yet when used from inside the
 callback. So I just switched the order there. There's no implication to
 the outside, it was just a bug caught by system tests, so I fixed it.

 > - not really for this branch, but `_add_remote_config_internal` seems
 >   to ignore some error cases:
 >   - non-0 rcode or value is None

 Being none is allowed, if there's no config set yet. But I do check the
 others now.

 > }}}
 > - error cases do not seem to be tested like this one:
 > {{{#!python
 >                 if module_spec.get_module_name() != module_name:
 >                     raise ModuleCCSessionError("Module name mismatch: "
 +
 >                                                module_name + " and " +
 >
 module_spec.get_module_name())
 > }}}
 >   (you may also want to run pycoverage)

 I tried running pycoverage, but it fails for me with this error:
 {{{
 Running test: edns_python_test.py
 Traceback (most recent call last):
   File
 "/home/vorner/work/bind10/src/lib/dns/python/tests/edns_python_test.py",
 line 18, in <module>
     from pydnspp import *
 ImportError: dynamic module does not define init function (initpydnspp)
 make[7]: *** [check-local] Error 1
 }}}

 This seems to happen reliably with all wrapper modules, and I need to look
 into it sometime. I didn't want to hold this ticket because of it, though.

 > '''changelog'''
 >
 > - s/tsig_kes/tsig_key_ring/
 > {{{
 > However, the old
 > configuration of Xfrout/tsig_kes need to be removed for Xfrout to
 > work.
 > }}}

 Hmm, yes, you're right. I'll update it when merging.

-- 
Ticket URL: <http://bind10.isc.org/ticket/1643#comment:9>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development