Dear Colleagues (and I do mean dear):
I think I will tell this story backwards. My original
question/problem was that I was switching my vnmr/vnmrJ instructional
server from a Sun Ultra-10 running Solaris (7, but I had security
hardened it to be as solid as 9, which I still love) to a Dell
Precision 390n running RHEL 4.0u3.
The Dell Windows XP PCs in the undergrad computer lab were using
PuTTY to ssh in and then get back vnmr or vnmrJ using X-Win32. The
Ultra-10 was doing fine with vnmr 6.1C, but, with only 640 MB RAM, it
could not accommodate more than 8 logins using vnmrJ (too much java
overhead). I tried switching to the Dell - and when it didn't work, I
sent out my plea for help.
Working from the solution backwards, I found that when X-Win32 was
upgraded to the most recent version, the scheme worked just fine from
both the linux and Solaris boxes. Rich Shoemaker has been through the
same thing and confirms this. This scheme means enabling
X11-forwarding in /etc/ssh/sshd_config, making sure ports 6000-60xx
are open on the receiving end, etc. It serves us well - fast, no glitches.
Many people were positively convinced that I needed to use the
RealVNC server/client software to do remote X-sessions. Everyone with
experience with both VNC and a similar product unanimously agreed
that "I would strongly recommend that you look at NX from NoMachine
(
http://nomachine.com/) as a much better alternative."
I can't comment on that. I have only enough time to put out fires -
the ssh tunneling is satisfying everyone and I understand it, so I am
sticking with it and expanding my instructional outreach to inorganic
chem lab and maybe instrumental analysis lab. (I am seeking job
security by putting my facility increasingly at the service of
curriculum enhancements. It seems like the faculty have fewer and
fewer $$$ for research spectrometer time. I've heard the same from
other colleagues at "modest" universities. Are you feeling the same
drop-off in research activity?)
Thanks so much to all of you. If I've left anything out or muffed
this summary, please write. Btw, the way I serve a large class like
undergrad organic chem is to use one account to transfer the data to
the standalone from the spectrometer. Then I move (for example
t-butanol.fid) to a student /vnmrsys/data directory as unknown32.fid
and delete all the other users' data directories, replacing them with
symbolic links to the first user's directory. I chmod as necessary to
keep the kids from spectrum-matching.
Do not let professors have the passwords to do this. Trust me. Try to
find the lock_1.primary in an account that has
~user/vnmrsys/data/vnmrsys/data/vnmrsys/data/vnmrsys/data/vnmrsys/data ... etc
Bill
William C. Stevens, Ph.D. Director
Nuclear Magnetic Resonance Facility
Southern Illinois University
Carbondale, IL 62901
618-453-6498 voice / -6408 fax / 521-9892 cell
http://opie.nmr.siu.edu
Received on Thu Jan 28 2010 - 17:41:03 MST