Age | Commit message (Collapse) | Author | Files | Lines |
|
irpc_servers_byname()
metze
(This used to be commit b54584dfabee77ec7743cab431bda9765057a295)
|
|
uint32_t server_id
to
struct server_id server_id;
which allows a server ID to have an node number. The node number will
be zero in non-clustered case. This is the most basic hook needed for
clustering, and ctdb.
(This used to be commit 2365abaa991d57d68c6ebe9be608e01c907102eb)
|
|
metze
(This used to be commit 8f9e201b9a797c0772672efab0f8e6a7a6312eb0)
|
|
rafal
(This used to be commit 48a9f822442c8b115fd61d9c6781d8100df2bf9e)
|
|
Andrew Bartlett
(This used to be commit e312cddafd7e00680dd059fad7ef7e5ecdbbb484)
|
|
try to include just the BASENAME.h files (containing only structs)
(This used to be commit 3dd477ca5147f28a962b8437e2611a8222d706bd)
|
|
file dependencies
(This used to be commit 122835876748a3eaf5e8d31ad1abddab9acb8781)
|
|
name, and use that.
(I was trying to find a machine by the name of __SAMBA__)
Andrew Bartlett
(This used to be commit cde044d023c7580442bceb60ac62dc4cfc1b85fe)
|
|
Andrew Bartlett
(This used to be commit 3e90e7edfa7d343a6b6bf073b8f4d018e3b463d0)
|
|
another case where we have to fallback to the node status request.
Andrew Bartlett
(This used to be commit 181064dbcf102de80937fc30b3d3ba5114194a72)
|
|
the remote server's name, or in the absence of a local nbt_server to
communicate with (or without root access), a node status request.
The result is that we are in a better position to use kerberos, as well
as to remove the 'password server' mandatory parameter for the samsync
and samdump commands. (I need this to put these into SWAT).
The only problem I have is that I must create a messaging context, which
requires a server ID. As a client process, I don't expect to get
messages, but it is currently required for replies, so I generate a
random() number. We probably need the servers to accept connections on
streamed sockets too, for client-only tasks that want IRPC.
Because I wanted to test this code, I have put the NET-API-* tests into
our test scripts, to ensure they pass and keep passing. They are good
frontends onto the libnet system, and I see no reason not to test them.
In doing so the NET-API-RPCCONNECT test was simplified to take a
binding string on the command line, removing duplicate code, and
testing the combinations in the scripts instead.
(I have done a bit of work on the list shares code in libnet_share.c
to make it pass 'make test')
In the future, I would like to extend the libcli/findds.c code (based
off volker's winbind/wb_async_helpers.c, which is why it shows up a bit
odd in the patch) to handle getting multiple name replies, sending a
getdc request to each in turn.
(posted to samba-technical for review, and I'll happily update with
any comments)
Andrew Bartlett
(This used to be commit 7ccddfd3515fc2c0d6f447c768ccbf7a220c3380)
|