BIND 10 #1790: update xfrin to have auth reload transfered zones

[BIND 10 Development]

#1790: update xfrin to have auth reload transfered zones
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:  muks
  jinmei                             |                Status:  reviewing
                       Type:  task   |             Milestone:
                   Priority:         |  Sprint-20120515
  medium                             |            Resolution:
                  Component:  xfrin  |             Sensitive:  0
                   Keywords:         |           Sub-Project:  DNS
            Defect Severity:  N/A    |  Estimated Difficulty:  3
Feature Depending on Ticket:  xfr    |           Total Hours:  0
  for in-memory                      |
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------
Changes (by vorner):

 * owner:  vorner => muks


Comment:

 Sorry for the late response.

 Replying to [comment:7 muks]:
 > > The handling of origin and dots is needlessly complicated. Could I
 propose creating Name objects of both and comparing them? This makes sure
 the comparison is correct one.
 >
 > Creating Name objects is indeed the right way. I've updated it to also
 use RRClass for the zone class comparison.

 There's one small problem in the changed code. We have a policy of using
 each log message identifier exactly once, so the place where it was logged
 can be easily identified by grep or something. But you use
 `XFRIN_AUTH_CONFIG_ERROR` at two places.

 > > The ignoring of errors when sending/receiving is suspicious at least.
 What is the purpose? And, if it was unable to send because the msgq
 terminated, then it should be FATAL error and terminate directly (the msgq
 must not quit). Anyway, I don't know if we want to handle this at all with
 try-catch, or maybe the receiving of answer.
 >
 > The msgq must not quit. I coded this based on another method in xfrin
 which sends out "notify" to xfrout. We can abort there, but ignoring the
 error (basically unable to ask auth to load zones) is not bad as long as
 auth can keep serving. If msgq dies, it will probably get restarted if it
 is of kind necessary?

 Well, msgq crashing is considered fatal and can not be changed. If msgq
 quits, the system goes down completely. That's why I think the catch is
 suspicious and should be removed (it is not properly handled error, so the
 error should not be handled at all).

 > > Why do you switch to IPv4 in the lettuce tests? I believe we had both
 types of addresses to make sure we work with both and if we should choose
 one, it should be the IPv6, not 4 I think.
 >
 > The test was configured with 127.0.0.1 in some files and ::1 in other.

 Not that it would be very important, but as I said, we want to make sure
 both v4 and v6 addresses work correctly, therefore we have a mix of them.

-- 
Ticket URL: <http://bind10.isc.org/ticket/1790#comment:9>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development