bind and short TLD problems?

[nexusis]

I'm new to ispCP - just upgraded a server from a RH5.2 install (where everything was manual and worked fine) and am now running ispCP 1.0.5 on Debian (Debian 4.1.2-25 according to /proc/version)

I have ~12 domains on this particular server, and all run fine except for 2... both are in the .CA TLD. However the registrars for the domains are different. The only commonality I can see is that they are .CA . I think its unlikely that the TLD itself is an issue - but perhaps the length is?

Is there a problem/configuration issue with short TLDs in the default install of ispCP 1.0.5?

The specific problem I'm seeing is that the dns entries will not resolve...

Bind is loading the zone - daemon.log shows the zone reloads and no errors are being thrown

Jul 15 01:55:45 apollo named[19272]: zone domain.ca/IN: loaded serial 2010071000
Jul 15 01:55:45 apollo named[19272]: zone domain.ca/IN: sending notifies (serial 2010071000)

but if I query the local name server I see errors

> server server.com
Default server: server.com
Address: ::ffff:a.b.c.d#53
Default server: server.com
Address: a.b.c.d#53
Default server: server.com
Address: a.b.c.d#53
> domain.ca
Server: server.com
Address: ::ffff:server.com#53

** server can't find domain.ca: NXDOMAIN

but queries for other domains on the host work just fine

> domain2.com
Server: server.com
Address: ::ffff:a.b.c.d#53

Name: domain2.com
Address: g.h.i.j
>

I suppose I could just replace the .db file with the one that I know was working previously (my rh5.2 also ran Bind9) - but I dont want to mess up how ispCP is managing things...

Any insights?