[ENBD] Two enbd-client on the same pc

[HUDZIA Bertrand]

		Hi every one ;)


My next problem is very simple, I didn't manage to set up a "mini" cluster:

I got 2 pc that each run nbd-server and a third pc that should run 2
nbd-client. Next step is to use soft-raid, but let talk about that first 
one ;)

If i do not set signature on the 2 server, as they are random generated,
the first client connect and "taint" the nbd module of the kernel. In
that case , the second nbd-client can't connect because he got a "false"
signature.

OK, let's try something else.

I set up the same signature on the both server. I connect the first
client, no problem. But when I connect the 2nd, i got strange result ,
once i manage to get 2 device ( /dev/nb2 and /dev/nb3 ) being the same
connection ( ie that they were both connected to the server #1 and the
#2 was offline but connected to ..?? ).
Well, that 's not possible that I wrote something wrong, the enbd server
watch for the ip and kill the old session if a new one arrive ( Am i
wright ?).
And the other times, i got this msg for the 2nd client :

	[root@bd1 root]# nbd-client bd3:1593 -n 1 -s /dev/nb3
	nbd-client   985: ok
	nbd-client   985: client (-1) manager opened NBD device /dev/nb0 (2b00)
	nbd-client   985: client (-1) left kernel bdflush sync boundary at 80%
	nbd-client   985: client (-1) left kernel bdflush async boundary at 25%
	nbd-client   986: client (-1) starts introduction sequence on port 1593
	nbd-client   986: client (-1) got session port 1594 ok
	nbd-client   986: client (-1) introduction sequence ends ok
	nbd-client   987: launch client (-1) childminder failed to open subdevice 1 for daemon (0)
	nbd-client   987: client (-1) last error No such file or directory
	nbd-client   986: sighandler relaunches child from manager 
	nbd-client   986: client (-1) childminder verifies pid 987 (0) is down
	nbd-client   986: client (-1) childminder launched pid 989 (0)
	nbd-client   989: launch client (-1) childminder failed to open subdevice 1 for daemon (0)
	


The 1st client's output was :

	[root@bd1 root]# nbd-client bd2:1592 -n 1 -s /dev/nb2 
	nbd-client   981: ok
	nbd-client   981: client (-1) manager opened NBD device /dev/nb0 (2b00)
	nbd-client   981: client (-1) left kernel bdflush sync boundary at 80%
	nbd-client   981: client (-1) left kernel bdflush async boundary at 25%
	nbd-client   982: client (-1) starts introduction sequence on port 1592
	nbd-client   982: client (-1) got session port 1593 ok
	nbd-client   982: client (-1) introduction sequence ends ok
	nbd-client   983: ok
	nbd-client   983: client (0) opened socket 5 to port 1593
	nbd-client   983: client (0) read passwd ok from port 1593
	nbd-client   983: client (0) got cliserv magic ok from port 1593
	nbd-client   983: client (0) got a signature ok from port 1593
	nbd-client   983: client (0) locking itself in memory
	nbd-client   983: client (0) begins main loop


Ok , i've just noticed one thing : the 2nd line of each client I ran is very distrubing :

	nbd-client   981: client (-1) manager opened NBD device /dev/nb0 (2b00)

They both say the want to run on /dev/nb0 ?? I ran once nbd-client with
/dev/nb10 and i got a "/dev/nb10" at the output .. err strange isn't it ?


Well , before kestion : The three pc ran the same install of linux :
RedHat 7.2 . They all compiled enbd 2.4.28 ;) They are all with 2.4.18 single
processor kernel.



		Bd.