Newsgroups: comp.parallel.pvm
From: clint@hal9000.unipv.it (Francesco Defilippo )
Subject: Pvm problems
Keywords: pvm
Organization: Pavia University
Date: 13 Jan 1995 10:31:45 GMT
Message-ID: <3f5kqh$i49@mirage.unipv.it>


We search an answer to these problems:

1)we run PVM3.3.6 sessions on a workstation HP/735, simulating the execution of an algorithm on a ring of 4 processors, but really using only one.
The elaboration ends correctly, but when we try to use XPVM to monitor times of execution and other parameters, the system crashes.
The reason of the crash is in the grewing use of the Ram (till to 60 MB); the 
operating system kills our processes when they use an amount of Ram higher than
a fixed limit.

2)alternatively we are running PVM3.3.6 on a cluster of workstations HP/705;
the cluster has been organized as follows: there is a server (named g32)
with two sons (g29, g30) and the other machines attached to g29 or g30.
All the workstations shared the file system on g32, but each one  works
independently from the others.
We try to run Pvm on two machines g25 and g26, assigning the two addresses to an
hostfile and typing 'pvm hostfile' on one of them. The daemon process starts on
only one machine and not on the other.
The two machines are connected because remsh commands work correctly.
We suspect that the problem could be originated by the log files created by one of the daemons we try to start. 
If the other daemon finds a pvml.log file with a PID different from its, it doesn't start correctly.
How is possible to resolve these problems? 
If somebody could help us, please answer me 

Francesco Leporati
Computer Science Department
University of Pavia-Italy

microlab@ipv85.unipv.it

