BIND 10 #2911: don't fallback from ixfr to axfr due to data source error

[BIND 10 Development]

#2911: don't fallback from ixfr to axfr due to data source error
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:
  jinmei                             |                Status:  new
                       Type:         |             Milestone:  Next-Sprint-
  defect                             |  Proposed
                   Priority:         |              Keywords:
  medium                             |             Sensitive:  0
                  Component:  xfrin  |           Sub-Project:  DNS
               CVSS Scoring:         |  Estimated Difficulty:  0
            Defect Severity:  N/A    |           Total Hours:  0
Feature Depending on Ticket:         |
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------
 See https://lists.isc.org/pipermail/bind10-dev/2013-April/004559.html

 It doesn't make sense to fall back from IXFR to AXFR if the reason
 for the failure is data source error.  Ideally, xfrin should be
 more resilient for temporary failure such as 'DB is locked', but
 that will be a little bit harder.

 For the moment I suggest just not to fallback in such cases.
 Specifically, we should explicitly catch `DataSourceError` type of
 exception in xfrin and handle it a non recoverable error for that
 xfrin session.  Optimistically, when zonemgr retries the transfer
 (maybe after several times) it will succeed.

 I also suggest revisiting which cases can actually be ixfr-to-axfr
 fallback in this ticket.  Check also the BIND 9 (or for that matter,
 any other, if you like) implementation.

-- 
Ticket URL: <https://bind10.isc.org/ticket/2911>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development