nbd questions and problems

nbd questions and problems

[haiquy]

Hi,

I have just tested nbd and got touble with it.

Kernel 2.6.8.1 (from kernel.org)

nbd server and client version

When I download the latest version there are many patches under the directory
nbd but when run configure it seems not to apply any patch. No README file
to instruct anything.
So I just ./configure then make. produce nbd-client and nbd-server
The nbd kernel module is the original one from 2.6.8.1

dd if=/dev/zero of=swap bs=1M count=64
nbd-server 1024 swap

The client is a diskless ; get its root via NFS . In client I run

nbd-client 10.0.0.2 1024 /dev/nbd/0

mkswap /dev/nbd/0

swapon /dev/nbd/0

It seems fine, but when the system touch swap there are error in the kernel log

nbd0 Receive control failed result -104

and the server exited .

If I run
nbd-client 10.0.0.2 1024 /dev/nbd/0 -swap

it ask me to apply some patches to work but there is no documentation I found on net
or in the nbd tar ball tell me which patch t apply

And now if I dont use swap, I did exactly like the above and then run
mke2fs /dev/nbd/0

It immiedately gave errors like
---
nbd0: Receive control failed (result -104)
nbd0: shutting down socket
-- many other error which is because of the above one

Is there anyway to fix these? please help.

Steve Kieu

Re: nbd questions and problems

[Paul Clements]

 > It seems fine, but when the system touch swap there are error in the
 > kernel log
 >
 > nbd0 Receive control failed result -104
 >
 > and the server exited .

Well, -104 is Connection reset by peer. Nothing in nbd would cause that 
directly. Without seeing the error messages from the nbd-server it's 
hard to tell, but I suspect that either nbd-server is dying at startup 
(check the syslog on the server, a common cause is failure to open the 
file/partition) or you have some sort of networking issue. Can you 
otherwise communicate between the two systems over a TCP connection ?

--
Paul

Re: nbd questions and problems

[Steve Kieu]

 --- Paul Clements <paul.clements@steeleye.com> wrote:

>  > It seems fine, but when the system touch swap
> there are error in the
>  > kernel log
>  >
>  > nbd0 Receive control failed result -104
>  >
>  > and the server exited .
> 
> Well, -104 is Connection reset by peer. Nothing in
> nbd would cause that 
> directly. Without seeing the error messages from the
> nbd-server it's 
> hard to tell, but I suspect that either nbd-server
> is dying at startup 

Nope, when I start the server it is running (waiting
for communication. When the client connect it is still
running (it print some thing like begin loop
request..). It only exited to the bash prompt without
any message. The syslog has a message (just found it)

nbd_server[16808]: Request too big!

probably is the reason why the server exited.

> (check the syslog on the server, a common cause is
> failure to open the 
> file/partition) or you have some sort of networking
> issue. Can you 
> otherwise communicate between the two systems over a
> TCP connection ?

I am sure the file is ok ; the communication is ok. As
later I succesfully ran enbd reliably (only with
2.4.27 in client, the 2.6.7 OOPs) so we can confirm
the TCP is good AFAIK enbd uses the same protocol. Can
ssh between the two machine, can http (run web browser
in client and access a page served by the server box
vise versa).

I will do more testing today. I need 2.6.8 kernel in
the client.

Kind regards,



=====
S.KIEU

Find local movie times and trailers on Yahoo! Movies.
http://au.movies.yahoo.com