[Wien] parallel ssh error

Gavin Abo gsabo at crimson.ua.edu
Sat Sep 28 08:06:01 CEST 2019


> Respected Sir, In my linux(Ubuntu 18.04 LTS) in ssh_config, and in 
> sshd_config there are two line already "SendEnv LANG LC_*" "AcceptEnv 
> LANG LC_*" respectively.

The "LANG LC_*" probably only puts just the local language variables in 
the remote environment.  Did you follow the previous advice [1] of 
trying to use "*" to put all variables from the local environment?

[1] 
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg19049.html

> However, ssh vsli1 'echo $WIENROOT' gives nothing (blank).

That seems to be the main cause of the problem as it should not return 
(blank) but needs to return "/servernode1" as you previously mentioned [2].

[2] 
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg19036.html

Perhaps the message below is a clue.  It you had set the WIENROOT 
variable in .bashrc of your /home/vlsi accounts on each system, you 
likely have to login and use that same /home/vlsi account on the head 
node as the output below seems to indicate login to a different 
/home/niel account. Alternatively, setting the WIENROOT variable in 
.bashrc of all /home/niel accounts on each node might work too.

>    The command ssh vsli1 'pwd $WIENROOT' print "/home/vlsi" the common 
> home directory and
> ssh vlsi1 "env"
> ...
> USER=niel
> PWD=/home/niel
> HOME=/home/niel
> ...
> this is similar as server, and other nodes.
>
> Sir After changing the parallel option file in $WIENROOT in server to
>
> setenv TASKSET *"yes" from "no"*
> if ( ! $?USE_REMOTE ) setenv USE_REMOTE 1
> if ( ! $?MPI_REMOTE ) setenv MPI_REMOTE 1
> setenv WIEN_GRANULARITY 1
> setenv DELAY 0.1
> setenv SLEEPY 1
> setenv WIEN_MPIRUN "mpirun -np _NP_ -machinefile _HOSTS_ _EXEC_"
> setenv CORES_PER_NODE 1
>
> the error is not coming but the program is not increasing steps after 
> lapw0 it stuck in lapw1

Since it seemed to be throwing an appropriate error message with TASKSET 
previously unlike when set to "yes", probably you should change it back 
to "no".

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20190928/e039c801/attachment.html>


More information about the Wien mailing list