View Full Version : [ubuntu] MPICH2: failed to handshake with mpd on node0

November 8th, 2010, 08:44 AM
Hello community,

I tried to setup a cluster using this howto: https://wiki.ubuntu.com/MpichCluster

What I have:
* nfs setup with same home directory for node and master
* password-less ssh connection
* mpich2 installed on both machines
* host entries in /etc/hosts(I can ping on hostnames)

But it fails, when I try to mpdboot on master(doesn't matter if mpd runs on the node or not). That's what I get:

master@master:~$ mpdboot -n 2 -v
running mpdallexit on master
LAUNCHED mpd on master via
RUNNING: mpd on master
LAUNCHED mpd on node0 via master
mpdboot_master (handle_mpd_output 407): failed to handshake with mpd on node0; recvd output={}I googled around, but I couldn't fint a working solution for me...

My /etc/hosts on master looks like: localhost master node0My /etc/hosts on node0: localhost node0 masterThe IP adresses are right, since I can ssh and ping the other machine using the hostname.

Thanks in advance

greetings, naeg

November 9th, 2010, 11:43 AM
Hi, I'm having the same problem.

meteo@boira:~$ cat mpd.hosts

meteo@boira:~$ mpdboot -n 3
mpdboot_boira (handle_mpd_output 407): failed to handshake with mpd on boira2; recvd output={}

The curious thing is that it runs fine for the master and one node
meteo@boira:~$ mpdboot -n 2
meteo@boira:~$ mpdtrace

and if I remove boira2 and leave only node boira3 then it runs fine.

Hope someone reads this message and can help.


November 10th, 2010, 09:51 PM

make your hosts:

master: localhost master node0

node0: localhost master node0