Dear List, Sorry if this is a naive question, but is there a known bug with phenix (or python?) on PAE kernels? I have looked on phenix-online.org, but did not find anything. This problem is independent of the phenix version (I have tried 1.4-92, 1.4-3, 1.3, and several releases of cci_apps) and occurs with almost all of the programs (most fail in the cci_apps test_apps script). The problem appears to be specific to the PAE kernels, since the same programs run fine in the non-PAE counterparts. I am running Fedora 10 with kernel versions 2.6.27.24-170.2.68/PAE and 2.6.27.21-170.2.56/PAE. Phenix programs fail under both PAE kernels, but are fine under the non-PAE kernels. Thanks in advance for any help you can provide. Best regards, Nick Silvaggi long int too large to convert Traceback (most recent call last): File "/usr/swr/cci_apps-07_25_1259/cci_apps_sources/elbow/elbow/command_line/builder.py", line 10, in <module> from elbow.chemistry.any_chemical_format_reader \ File "/usr/swr/cci_apps/cci_apps_sources/elbow/elbow/chemistry/any_chemical_format_reader.py", line 3, in <module> File "/usr/swr/cci_apps/cci_apps_sources/elbow/elbow/chemistry/SimpleMoleculeClass.py", line 16, in <module> File "/usr/swr/cci_apps/cci_apps_sources/elbow/elbow/utilities/monomer_utilities.py", line 2, in <module> File "/usr/swr/cci_apps-07_25_1259/cci_apps_sources/elbow/elbow/command_line/__init__.py", line 4, in <module> File "/usr/swr/cci_apps-07_25_1259/cci_apps_sources/elbow/elbow/command_line/__init__.py", line 3, in <module> File "/usr/swr/cci_apps/cci_apps_sources/mmtbx/mmtbx/refinement/print_statistics.py", line 2, in <module> File "/usr/swr/cci_apps-07_25_1259/cci_apps_sources/elbow/elbow/command_line/__init__.py", line 1, in <module> # packaging File "/usr/swr/cci_apps-07_25_1259/cci_apps_sources/elbow/elbow/command_line/__init__.py", line 518, in <module> File "/usr/swr/cci_apps/cci_apps_sources/boost_adaptbx/boost/python.py", line 68, in __init__ OverflowError: long int too large to convert
On Jun 12, 2009, at 6:45 AM, [email protected] wrote:
Sorry if this is a naive question, but is there a known bug with phenix (or python?) on PAE kernels? I have looked on phenix-online.org, but did not find anything. This problem is independent of the phenix version (I have tried 1.4-92, 1.4-3, 1.3, and several releases of cci_apps) and occurs with almost all of the programs (most fail in the cci_apps test_apps script). The problem appears to be specific to the PAE kernels, since the same programs run fine in the non-PAE counterparts. I am running Fedora 10 with kernel versions 2.6.27.24-170.2.68/PAE and 2.6.27.21-170.2.56/PAE. Phenix programs fail under both PAE kernels, but are fine under the non-PAE kernels. Thanks in advance for any help you can provide.
OverflowError: long int too large to convert
I believe we've seen this error before with a pharma user, but we weren't able to get much detail about the system - in retrospect, it was probably a PAE kernel as well. Anyway, it's some kind of binary incompatability, so the best solution in this case is probably to compile from source. (I'll send you the link for that separately.) One question: is the PAE kernel the default or did you have to custom- compile it? I wasn't familiar with the term until now, and I haven't seen a kernel with that extension before. What is the result when you enter the command 'uname -a'? -Nat ------------------- Nathaniel Echols Lawrence Berkeley Lab 510-486-5136 [email protected]
Sorry if this is a naive question, but is there a known bug with phenix (or python?) on PAE kernels?
Thanks for the "PAE" clue! This finally sheds some light on the strange error you get, which I've seen a few times before. In my experience, even compiling from sources (including Python) doesn't help. I'm guessing the error will need to be fixed in Python. This may not happen quickly. Can you reboot your system without PAE? Starting with Fedora 11, which just came out this week, 64-bit kernels are the default even on 32-bit machines. IMO that's the best way forward. (The Fedora 11 system we have has been stable for the few days it is up and runs all the phenix tests without problems.) Ralf
To make things even stranger, I tried running the binary distribution of cci_apps on another machine with a PAE kernel and it works! Here is the uname output: Linux fluffy.bu.edu 2.6.27.21-170.2.56.fc10.i686.PAE #1 SMP Mon Mar 23 23:24:26 EDT 2009 i686 i686 i386 GNU/Linux Now, phenix fails under this same kernel on *my* machine, but on fluffy here it works fine. The only thing I can think of is that the problem is due to the hardware differences between the two: Fluffy has 2 quad core Xeons with DDR2 (I think) and my machine has Core i7 with DDR3. I don't know if this bit of detail helps or not, but there it is. I will try recompiling on my core i7 box this evening and let you know how it turns out. Cheers, Nick Ralf W. Grosse-Kunstleve wrote:
Sorry if this is a naive question, but is there a known bug with phenix (or python?) on PAE kernels?
Thanks for the "PAE" clue! This finally sheds some light on the strange error you get, which I've seen a few times before. In my experience, even compiling from sources (including Python) doesn't help. I'm guessing the error will need to be fixed in Python. This may not happen quickly. Can you reboot your system without PAE? Starting with Fedora 11, which just came out this week, 64-bit kernels are the default even on 32-bit machines. IMO that's the best way forward. (The Fedora 11 system we have has been stable for the few days it is up and runs all the phenix tests without problems.)
Ralf _______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
Not surprisingly, recompiling from source did not fix my problem.
Phenix (and python) run fine on several computers with PAE kernels in
the lab. So, it seems to me that this is likely a processor/chipset
issue more than a kernel problem. The only differences between the
machines that work and the one that doesn't is the hardware.
Quoting "Ralf W. Grosse-Kunstleve"
Sorry if this is a naive question, but is there a known bug with phenix (or python?) on PAE kernels?
Thanks for the "PAE" clue! This finally sheds some light on the strange error you get, which I've seen a few times before. In my experience, even compiling from sources (including Python) doesn't help. I'm guessing the error will need to be fixed in Python. This may not happen quickly. Can you reboot your system without PAE? Starting with Fedora 11, which just came out this week, 64-bit kernels are the default even on 32-bit machines. IMO that's the best way forward. (The Fedora 11 system we have has been stable for the few days it is up and runs all the phenix tests without problems.)
Ralf _______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
Not surprisingly, recompiling from source did not fix my problem. Phenix (and python) run fine on several computers with PAE kernels in the lab. So, it seems to me that this is likely a processor/chipset issue more than a kernel problem. The only differences between the machines that work and the one that doesn't is the hardware.
I'm surprised. My recommnendation is to install a 64-bit OS. Or do you have important reasonsons for preferring 32-bit? Ralf
participants (4)
-
Nathaniel Echols -
Nicholas R. Silvaggi -
nick＠silvaggi.com
-
Ralf W. Grosse-Kunstleve