***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***


dear colleagues,

when processing with mosflm626 in batch mode (but not when processing
interactively) we very frequently get mosflm crashes with 'system 5 i/o'
errors. the problem is somewhat elusive, since the crashes are not
strictly reproducible. when i process the same dataset using the same
parameters 10 times,  it may crash in seven attempts and run fine in
three. the problem also is not reproducible on all hardware. mosflm
gives no further details as to the i/o unit that causes the problem. the
crashes exclusively occur in problematic datasets and can almost always
be overcome by subtle changes in the processing parameters. while i can
automate the process of parameter adjustment, this behaviour consumes
more computing resources than desirable, and i was wondering whether
anybody would have observed this and would have an idea (short of buying
new hardware or installing a different operating system) about how to
catch these crashes cleanly.

we are using mandrake 10.2 with linux 2.6.11-12mdksmp and i have
ccp4-6.0.1 installed. the problem occurs with mosflm from this
distribution, but also with previous as well as the very latest release
from harry powell`s pages.

any ideas welcome ...

greetings

ingo

-- 
Ingo P. Korndoerfer, PhD
Head of Crystallography
CrystaX Pharmaceuticals, S.L.
Parc Científic de Barcelona 
Josep Samitier 1-5 
08028 Barcelona
Tel.: +34 93 403 9093
Fax: +34 93 403 4788
e-mail: [EMAIL PROTECTED]


Reply via email to