BIND 10 #2832: extend data source config to specify shmem segment params

[BIND 10 Development]

#2832: extend data source config to specify shmem segment params
-------------------------------------+-------------------------------------
            Reporter:  jinmei        |                        Owner:
                Type:  task          |                       Status:  new
            Priority:  medium        |                    Milestone:  Next-
           Component:  Unclassified  |  Sprint-Proposed
            Keywords:                |                   Resolution:
           Sensitive:  0             |                 CVSS Scoring:
         Sub-Project:  DNS           |              Defect Severity:  N/A
Estimated Difficulty:  0             |  Feature Depending on Ticket:
         Total Hours:  0             |  shared memory data source
                                     |          Add Hours to Ticket:  0
                                     |                    Internal?:  0
-------------------------------------+-------------------------------------

Comment (by vorner):

 Hello

 Replying to [comment:2 jinmei]:
 > > Shouldn't that be better a global memmgr setting?
 >
 > Where to specify this information is actually something I'm not sure
 > yet, but I'm also not sure what the "global" memmgr setting is.  Can
 > you show an example?

 Well, outside of the `data_sources` pseudo-module. Something like
 `/MemMgr/storage "/path/to/directory"`. It would default to something near
 the
 place we store config and sqlite databases. Then it would simply create
 files
 as it needs there and tell the other modules in which file it'll find
 which
 zone.

 Because, in your case, what happens if you have one data source with one
 setting and other with different? Does it choose one? Does it create a
 separate
 file for each of the data sources? If so, will it be able to comply with
 setting both data sources to the same value correctly?

 > After thinking about this a bit closely, I'm currently feeling that it
 > makes sense to simply specify a file (with postfix for versioning)
 > just like the way we specify the sqlite3 DB file.  But as I said
 > this is still a bit open.

 If there's more than one file (like with the versioning), I think it is
 more
 intuitive to use directory. If I set a filename, I expect that file to
 exist
 and not other with some suffix, or multiple files. If I set directory, I
 expect
 the program to do whatever it likes in the directory.

-- 
Ticket URL: <http://bind10.isc.org/ticket/2832#comment:4>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development