|
RE: .0
When you issue "MCC> dir node4 * " command, the entire DNS namespace (every
directory) is searched by DNS Clerk in your system.
If any one directory in the namespace is not accessible, you may get the
%DNS-E-NOCOMMUNICATION error.
If a directory (either master or read-only copy) resides in a remote
clearinghouse, and the namesever for the clearinghouse is not reachable,
DNS clerk returns %DNS-E-NOCOMMUNICATION status.
One way of checking that every directory is reachable is to use an example
program which displays sub-directories of a directory of a namespace, and all
objects stored in them. The example (DNS_OBJ_WALK.C) is in Topic # 34.5 in
NOTED::DNS_PROGRAMMER notes conference. (Note 34.8 tells how to link the
program) I think that if a directory is not reachable, the program may
display the %DNS-E-NOCOMMUNICATION status.
Another way of checking that every directory is reachable is to use DNS$Control
program. If your namespace has a small number of directories, you can use
DNS$Control command for each directory:
$ mcr dns$control
DNS> sho dir .dna_node known object
DNS> sho dir .mydir known object
.
.
.
.
>> In EVERY case, MCC performs a successful directory look up until it
>> reaches the SECOND TO LAST entity instance, at which point I get this:
The reason for the lookup operation failure showing up at the SECOND TO LAST
instance is that MCC does have a look-ahead buffer. After the last instance
is read from the DNS Clerk (meaning the second to last instance is returned to
the caller, and the last instance is in the look-ahead buffer), DNS Clerk
fails to access a directory, the error status is passed to the caller.
William
|