-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi David--

I'm forwarding this to the list, because it may bite others as well.

>
>> Hi Brady, the job I started yesterday morning wrote out all my
>> structure files and then hung up with messages in the .out.0 file
>> of "trying .-barrier-XXXX-X" and "waiting for process X", where X
>> are numbers I do not remember, and "waking all processes".
>>

These are normal messages for xplor-nih-2.14.

>> The next job I ran went through only half the calculations and
>> wrote out half the files.  It hung up after it calculated the first
>> batch of files in the "structureLoop" of my python script.  The
>> messages in the .out.0 file are "trying .-barrier-8064-1", "waiting
>> for process 1".  The other .out files I looked at simply look
>> suspended.

This (mis)behavior is known to occur. It probably means there are some
inconsistencies reading/writing files from/to the cluster's common
(probably NFS) file system. Due to problems like these, xplor-nih-2.15
uses a different synchronization mechanism.

The temporary workaround is to disable the statistics-gathering and
averaging features of StructureLoop: i.e. do not specify the
averageFilename, or genViolationStats parameters.

regards--
Charles
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Processed by Mailcrypt 3.5.8 <http://mailcrypt.sourceforge.net/>

iD8DBQFEvocOPK2zrJwS/lYRArexAKCF+4NeFux7gNoeskzdT+5L9z9ClgCeM+NP
abgFQWYcJQQP9TmDCN5+/bQ=
=t3OU
-----END PGP SIGNATURE-----

Reply via email to