Hi James,
There is no overall Phenix temp directory specification, but most of the
temp_dir usage is from autosol/autobuild/ligandfit/map_to_model. Each of
these has the keyword "temp_dir=xxxx" which you should be
able to set to any directory you want (and local is better as you note).
Most programs using a temp_dir also have a keyword clean_up=True as well.
All the best,
Tom T
On Sat, Apr 24, 2021 at 10:51 AM James Holton
Thank you Li-Wei
Definitely not placing blame on one program. Phenix.autobuild is another big temp file producer. So is XDS. Clearly this ligand run was a case of a misconfigured, runaway task that never finished. However, the files lingered on disk, eating up inodes for 3 years!
The reason I'm asking is I think there are significant performance increases to be gained by using fast, local storage for scratch files. This is not just in speed but storage and overall system/cluster performance. Very few things are more expensive than an NFS write!
Does anyone know how to change the default temp file location across phenix ? Is this a cctbx thing?
Thanks
-James
On 4/23/2021 9:38 PM, Li-Wei Hung wrote:
Hi James,
I'll leave the global Phenix temp aspect to Billy. For ligand identification specifically, the working directory is where all the files are located. The program will purge most of the intermediate files upon completion. If the user interrupted the runs or if the program crashed at certain spots, the purge mechanism might not kick in. Even so, it'd take many runs to accumulate 20e6 (2e7?) files. In any case, you've got a point and I'll look into salvaging intermediate files of ligand identification as soon as they are not needed in the process.
Thanks,
Li-Wei
On 4/23/2021 7:03 PM, James Holton wrote:
Hello all,
Is there a way to configure phenix at install time (or perhaps post-install) to put temporary files under /tmp ? I just had to delete 20e6 temp files over NFS from a single user's phenix ligand identification run. The delete took almost a month.
Apologies if I am neglecting to look somewhere obvious in the documentation,
Happy Weekend!
-James Holton MAD Scientist
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]
-- Thomas C Terwilliger Laboratory Fellow, Los Alamos National Laboratory Senior Scientist, New Mexico Consortium 100 Entrada Dr, Los Alamos, NM 87544 Email: [email protected] Tel: 505-431-0010