I have just gone through one of those 2-3 week software-hell
sessions, and have a couple suggestions/questions for you.
[Although the problems detailed are VNMR specific, the
questions and comments are general.]
The short version of what happened to me is the following:
1. User has a simple (stupid!!) password; hacker breaks it and
starts running irc's (internet remote chat) on all our Sun's.
We have to rebuild Suns from scratch, and decide to upgrade
to Solaris 2.6 and VNMR 6.1A (we had been at 2.5.1 and 5.3b).
2. All goes ok with 3 datastations and our INOVA Suns; last
is our UNITY. I call Varian specifically to ask if there are
any known problems with VNMR 6.1A on UNITY spectrometers; I
worry because I believe Varian is not well equipped any more
to test older hardware with the new software releases. I am
assured by Varian (at a high level in software personnel)
that there is no problem. I see some bug reports at Varian's
website, but assume patches have fixed any serious ones.
3. Bottom line is after nearly two weeks of constant trouble on
the UNITY, I back down the software on that Sun to 2.5.1 and
5.3B. Consultation with Varian and some other UNITY/UNITY+
users has me convinced that VNMR 6.1A will not work stably
on _any_ UNITY or UNITY+ spectrometer (Varian has not confirmed
or denied this statement which they've already seen from me).
Amazingly (or perhaps not, depending on how jaded you are),
the problem is caused by a software patch (one of those fixing
bugs, but creating new ones!), and Varian has known about it
since last summer. In fact, they have fixed the bug but only
in the not yet released VNMR 6.1B. They do not fix the 6.1A
software patch.
Well, I loss a significant number of hours (I heavily modify VNMR,
so the upgrades--and the downgrade--are nontrivial). More
importantly, we lost nearly two weeks of spectrometer time on
a 100% use 500 MHz spectrometer.
So my suggestions are:
1. I believe AMMRL (or more appropriately varnet, bruker_users_mail,
etc.) should be posted anytime a significant bug or hardware
problem is identified. One certainly would hope the vendors
would do this (see next), but it seems user input is desirable.
2. [Better on varnet, but I'm already here] Any UNITY/UNITY+
users, please post directly on varnet your experiences with
VNMR 6.1B coming hopefully next month. If any of you are
running 6.1A and seem to be ok, please let me know.
Questions are:
1. None of us want to get inundated with email. How do we determine
when "significant" is sufficient to post? Are bug reports (Varian
does post these) sufficient? My opinion is they are not, as they
do not contain sufficient information to determine how serious they
are. Moreover, it appears vendors are often _reluctant_ to
admitting serious bugs or hardware problems.
2. How many of you use ssh (secure shell) to avoid telnet/ftp open
password transfers? How many have had hacker trouble recently?
How many academicians have put firewalls in place? I guess the
main question is, how are various people responding to hacker
problems in UNIX, especially those that don't have firewalls?
We have had huge trouble at UW (especially with Linex, but recently
with Suns) over the last 6 months. I can send summaries of
responses to just this set of questions.
3. Last, how do we as a group get vendors to be more sensitive to
spectrometer downtime? I particularly get annoyed at installers,
who seem to think they can work at their leisure in getting the
system going. I certainly found deaf ears when I passed along
faculty comments/pressure to Varian and now Bruker installation
groups. One final comment on this point: if you can avoid it,
do not allow remote installs. We did this with Varian, and we
were at best 50% productive with these sessions. Usually,
Varian would not connect until around noon, and would work late
but often with slow or poor connections, so they did not get
nearly as much done as they would have on location. Our
estimate was we lost at least one week of spectrometer time, and
more likely two. Moreover, at the end I had many questions that
just didn't get answered nearly as well as they would have if
the engineer had been sitting next to me, rather than on a phone
or via email.
Hackers are hell,
----------------------------------------------------------
Charlie Fry Tel: (608)262-3182
Director, MR Facility Fax: (608)262-0381
Chem. Dept., Univ. Wisconsin
Madison, WI 53706 USA email: fry@chem.wisc.edu
----------------------------------------------------------