BIND 10 #3384: Manual testing of Kea for 0.8 (BIND10 1.2) release.

[BIND 10 Development]

#3384: Manual testing of Kea for 0.8 (BIND10 1.2) release.
-------------------------------------+-------------------------------------
            Reporter:  marcin        |                        Owner:
                Type:  task          |  marcin
            Priority:  high          |                       Status:
           Component:  dhcp          |  accepted
            Keywords:                |                    Milestone:  DHCP-
           Sensitive:  0             |  Kea0.9-alpha
         Sub-Project:  DHCP          |                   Resolution:
Estimated Difficulty:  0             |                 CVSS Scoring:
         Total Hours:  0             |              Defect Severity:  N/A
                                     |  Feature Depending on Ticket:
                                     |          Add Hours to Ticket:  0
                                     |                    Internal?:  0
-------------------------------------+-------------------------------------

Comment (by marcin):

 In addition to testing that Tom has performed. I did the following:

 Go over the steps to setup PostgreSQL database for Kea:
 - Database is setup without errors
 - Kea can use database

 Issue:
    - Some of the SQL commands should not be terminated with a semi colon.
 For example:
 {{{
 postgres-# GRANT ALL PRIVILEGES ON DATABASE kea to kea;
 ERROR:  syntax error at or near "kea"
 LINE 1: CREATE USER kea WITH PASSWORD kea
 }}}

   while it wouldn't fail if the semi colon was not appended at the end of
 the SQL line. Tom   said he will correct the examples in the bind10-guide
 to not include semicolon where they must not appear.

 ----

 Subnet configuration
 Outcome:
 - Server doesn't allocate address until subnet is configured
 - Server doesn't respond to queries when configured subnet is not matching
 IPv4 address on the specific interface
 - Server responds to queries when subnet is configured and it matches the
 address on the interface on which the query has been received
 - It should be not possible to create two subnets having the same subnet
 id. The first subnet id set to 0, and the second one set to 1 should cause
 a configuration error because for the first subnet the id = 1 is
 autogenerated.

 ----

 Memfile backend
 Outcome:
 - Default configuration should trigger use of the Memfile backend in
 persistent mode and leases should be written to file in a default location
 - It should be possible to disable the persistence and server should stop
 writing leases to a file
 - It should be possible to reenable persistence and server again writes
 leases to a file
 - It should be possible to change the lease file location and the lease
 file should be created

 Issue:
 - There is a known problem with the bind10 configuration mechanism. This
 shows up when partial configuration is committed for the map element. When
 second commit comes, after additional changes, the previous changes get
 lost. So, it is possible to select the Memfile backend by simply setting
 the type to !''memfile!''. If I want to later change the location of the
 lease file and I set the name parameter and commit, the configuration will
 fail because the type is lost. This means that all parameters have to be
 always specified on each reconfiguration. Perhaps this should be noted in
 the release notes?

 ----

 Switching from Memfile to Postgres
 Outcome:
 - When server is running and Memfile is used, it should be possible to
 change the backend to Postgres (online reconfiguration) and the new
 backend should be used (leases should be written to the database).
 - It should be possible to revert to the Memfile backend and observe
 leases being written.

 ----

 Renew
 Outcome:
 - With short T1 time we should observe clients renew their leases and
 database should be updated accordingly

 ----

 DHCPv6 specific: Specify both address and prefix pools
 Outcome:
 - Client should be able to request IA_NA only
 - Client should be able to request IA_PD only
 - Client should be able to request both IA_NA and IA_PD
 - In lease database or file it should be a distinction between lease type
 NA and PD.

 Known issues:
 - Memfile backend doesn't differentiate between IA_NA and IA_PD so odd
 things happen when client is trying to get IA_NA and IA_PD in the same
 request. Basically, he will get both address and prefix, and
 address=prefix. See #3148. But, should we mention this in release notes?

 Issue:
 - There is no pool overlapping detection so it is possible to create
 overlapping pools for IA_NA and IA_PD. The same client will get the
 address and prefix, and the address will belong to the prefix he gets.
 Also, it is possible that other client will get address which belongs to
 the prefix that someone else gets.


 All of these (except the first issue) can be addressed after a release.

-- 
Ticket URL: <http://bind10.isc.org/ticket/3384#comment:3>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development