Hi -
Yes, your suspicion is correct, and I think that bug has been silently
fixed in the latest distribution.
I am pretty sure Victor has now a job-dependent filename, if not we
will try and fix it now as we prepare
a new release. In that case, pleas email Gerrit telling him details
about your platform, and he
can send you a new executable hmain and instructions for how to fix
this till then.
best,
Tassos
On Mar 24, 2011, at 16:56, Matthew Franklin wrote:
Hi all -
I'm finding that Arp/Warp is crashing when I try to run a job. I
was working on a challenging molecular replacement job where I
though Arp tracing the fragmentary density could do a better job
than I could by hand (this turned out to be true, by the way). I
was testing different parameters, and running a few flex-warp jobs
in parallel. They would crash at different cycles of the building
and refinement, but always at the same point in the cycle. Looking
at the error messages in the log files led to this one, in the last
HmainPept log file to be written:
Density interpolation: points 10539450
0.00 CPU 2 Elapsed Superposition
11394 rotations have been done, total 46821
0.00 CPU 1 Elapsed Rotations
2009 flipped peptides are taken
end_tag_cputime
=========================
data send to stderr:
forrtl: severe (64): input conversion error, unit 15, file /Users/
matt_franklin/bin/arp_warp_7.1/fort.15
Image PC Routine Line Source
hmain 00106F9F Unknown Unknown Unknown
hmain 001063CB Unknown Unknown Unknown
hmain 000CCD30 Unknown Unknown Unknown
hmain 0009217B Unknown Unknown Unknown
hmain 0009192A Unknown Unknown Unknown
hmain 000B35C8 Unknown Unknown Unknown
hmain 000B0C79 Unknown Unknown Unknown
hmain 00012A35 Unknown Unknown Unknown
real 11.60
user 11.51
sys 0.08
I think what's happening here is that I had multiple jobs running
simultaneously, and they were all trying to read and write to the
same fort.15 file. As you can see, this file is in Arp/Warp's main
directory, not in any of the working directories. At some random
point, one job would step on the other's read or write, and I'd get
a crash.
Can someone confirm my guess? Is it possible to fix this bug so
that multiple flex-wARP jobs can be run in parallel? I'm running
"Arp/Warp Expert System" (i.e. flex-wARP) from ccp4i, and I don't
think this is happening with other Arp/Warp modes.
(I'll be happy to provide more logfiles, input files, etc.)
Thanks,
Matt
--
Matthew Franklin, Ph. D.
Senior Research Scientist
New York Structural Biology Center
89 Convent Avenue, New York, NY 10027
(646) 275-7165
P please don't print this e-mail unless you really need to
Anastassis (Tassos) Perrakis, Principal Investigator / Staff Member
Department of Biochemistry (B8)
Netherlands Cancer Institute,
Dept. B8, 1066 CX Amsterdam, The Netherlands
Tel: +31 20 512 1951 Fax: +31 20 512 1954 Mobile / SMS: +31 6 28 597791
