[ENBD] (no subject)

[Peter T. Breuer]

"A month of sundays ago HUDZIA Bertrand wrote:"
> 
> The point is that i make test for all that could suit. Unfortunatly, i ran iozone on enbd-2.4.28 and i got on the server side :
> 
> nbd-server  1416: newproto Not enough magic in packet. Breaking off.

Network borked or client died.

> nbd-server  1416: slavesighandler server (0) activates slave sighandler for signal 15
> nbd-server  1416: server (0) sighandler terminates slave 1416 safely
> nbd-server  1415: server (-1) relaunches child after SIGCHLD 

Well, the server died. That's correct.

> nbd-server  1415: server (-1) slave pid 1416 is down, launching new
> nbd-server  1415: server (-1) launched slave pid 1430
> 
> The server still ran but the client is just out, hard reboot is necessary.

No it isn't. Try "echo 0 > /proc/nbdinfo".

> Both ran 2.4.18 kernel under a RedHat-7.2 , the shared device was 35Go in reiserfs (problem also on ext2).

Shared device? That makes me suspicious. Nothing should be shared!
It's borrowed :-).

> The number ofchannel doesn't change anything. I didn't put any special option for both client and serv .
> 
> I didn't find out what was going on .. 
> 
> But i ran an other test and crashed the client again : 
> 	tar -xzf linux-2.4.18.tar.gz &
> 	cd linux.two && make bzImage -j 10
> 
> Can anyone can help me ? I'll be sorry to put aboard enbd ;)

I can't tell from here. But the problem seems clientside. What kind of 
kernel and architecture are you running? Are you sure you have compiled
consistently .. i.e. all module, server and client from the same
package?

Let me know and I'll comment further. I also need to see /proc/nbdinfo.

Peter