Running Charmrun/NAMD with more than 10 processes fails

From: K Miura (080m812m_at_stu.kobe-u.ac.jp)
Date: Tue Dec 16 2014 - 02:53:36 CST

I am using NAMD 2.10 on two MacPro 2014 nodes with MacOSX 10.9.
I downloaded MacOSX-x86_64-netlrts binary from the NAMD web site, and
put the program in /usr/local/bin/ of each computer.
When I run the NAMD with two MAC computers by using the command below:

  charmrun +p11 ++verbose /usr/local/bin/namd2.10 ++remote-shell ssh
NAMD.z.cfg

Then Charmrun program stops with log output below.

Charmrun>charmrun started...
Charmrun>using ./nodelist as nodesfile
Charmrun>adding client 0: "192.168.x.49", IP:192.168.x.49
Charmrun>adding client 1: "192.168.x.50", IP:192.168.x.50
Charmrun>adding client 2: "192.168.x.49", IP:192.168.x.49
..
Charmrun>adding client 13: "192.168.x.50", IP:192.168.x.50
Charmrun>adding client 14: "192.168.x.49", IP:192.168.x.49
start_nodes_rsh
Charmrun> Sending "0 192.168.x.45 52597 8206 0" to client 0.
..
Charmrun> clisnt 1 connectd (IP=192.168.x.50) data_port=54928)
Charmrun> clisnt 13 connectd (IP=192.168.x.50) data_port=55722)
Charmrun> clisnt 11 connectd (IP=192.168.x.50) data_port=50657)
Charmrun> Waiting for 13-th client to connect.
Timeout waiting for node-program to connect

When I run the charmrun with +p10 or lower, the namd runs successfully
in each computer node.
Running the same program with ++local option failed calculation:

  charmrun ++local +p11 ++verbose /usr/local/bin/namd2.10
++remote-shell ssh +idlepoll NAMD.z.cfg

I suspected my misconfiguration of ssh server/client, so I googled the
term "ssh max connections"
but could not get helpful hits.
I also searched namd-l logs, NAMD tutorial and FAQs with the error
message "Timeout waiting for" etc., but I could not find any solution.

my nodelist file is:

group main
host 192.168.x.49
host 192.168.x.50

host 192.168.x.49 and 192.168.x.50 allow ssh login by using public key
without password prompt.

Setting of max connections in /etc/sshd_config are below:
MacSessions 50
MaxStartups 50:30:100

Could anyone give me ideas?

Thank you in advance.

-----------------------------
Kenji Miura
Kobe University Graduate school of Medicine

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:23:07 CST