Fedora 27 has just been released in final form, I did not expect to find any previous references to it yet. (I'm using x86_64, of course.) I preferred to use Fedora because I had little trouble getting it to hibernate, which might be important when using this software on a primitive computer. In the event of a power failure, although not frequent, the job would not be terminated by the event. The only problem I had encountered with Fedora 26 was with openmpi. Fedora does not put it into the standard search path, but into /usr/lib64/openmpi/bin, which must be added. I upgraded Fedora 26 to 27 in place, leaving the Cactus installation untouched. The only problem I had with it was that Fedora 27 removed /usr/lib64/libgsl.so.19, which I had to add after the upgrade was completed. Then I attempted to install Cactus on a freshly-installed Fedora 27 installation. Things went well for a while, but eventually I encountered this: COMPILING arrangements/McLachlan/ML_BSSN/src/RegisterSymmetries.ccCOMPILING arrangements/McLachlan/ML_BSSN/src/Startup.ccCreating /home/jontester/Cactus/configs/sim/lib/libthorn_ML_BSSN.amake: *** [Makefile:260: sim] Error 2 I would be interested to know why this occurred. For the present time it does not inconvenience me, but I'm sure the software will continue to evolve. (My name isn't Jon Tester, but I use this user for experimental software.)
Hello,
while, but eventually I encountered this: COMPILING arrangements/McLachlan/ML_BSSN/src/RegisterSymmetries.ccCOMPILING arrangements/McLachlan/ML_BSSN/src/Startup.ccCreating /home/jontester/Cactus/configs/sim/lib/libthorn_ML_BSSN.amake: *** [Makefile:260: sim] Error 2 I would be interested to know why
Hmm that error itself does not tell much. Make returns "2" for eg dependency errors and possibly other things. I would try compile again and see if there is any actual error message shown.
As for Fedora moving files and using "non-standard" locations, yes this is a problem we are struggling with. Thankfully on most workstations things work fine as long as pkgconfig is installed and (on fedora) one uses "module load mpi" before compiling.
Please have a look at:
https://docs.einsteintoolkit.org/et-docs/Simplified_Tutorial_for_New_Users
in particular at the Prerequisites and the FC25 settings:
su -c ' yum -y install mpich2 python mpich2-devel gsl gsl-devel libjpeg-devel hdf5 hdf5-mpich-devel gcc gcc-c++ gcc-gfortran patch numactl-devel numactl hwloc subversion git openssl-devel lapack-static'
which admittedly use mpich and not openmpi.
In addition I would strongly advise to make sure pkg-config is installed (often gets pulled in by default):
su -c 'ym install pkg-config'
Then
module load mpi ./simfactory/bin/sim setup-silent --optionlist=fedora.cfg --runscript debian.sh
and see if this lets you compile.
Yours, Roland
On Tue, Nov 21, 2017 at 04:33:14AM +0000, iconoklastic91@gmail.com wrote:
COMPILING arrangements/McLachlan/ML_BSSN/src/RegisterSymmetries.ccCOMPILING arrangements/McLachlan/ML_BSSN/src/Startup.ccCreating /home/jontester/Cactus/configs/sim/lib/libthorn_ML_BSSN.amake: *** [Makefile:260: sim] Error 2
Hi,
Different parts of Cactus are by default built in parallel, so a short excerpt of the build log might not be enough. We might be able to tell you more if you would log all of the output and send us that (if you know how, please compress it, it might be quite a bit).
If one of the external libraries didn't build (not unlikely), then most often that log only refers to the log of that particular library (usually called build.log). If that is the case, that file would be of particular interest.
Frank
I discovered that I was also missing environment-modules, and that might account for failure to load the module (so I'm going to try that, and see if it solves the problem). If it doesn't, I'll try to collect the output you mentioned from the failed build. On Monday, November 20, 2017, 11:37:11 PM PST, Frank Loeffler knarf@cct.lsu.edu wrote:
On Tue, Nov 21, 2017 at 04:33:14AM +0000, iconoklastic91@gmail.com wrote:
COMPILING arrangements/McLachlan/ML_BSSN/src/RegisterSymmetries.ccCOMPILING arrangements/McLachlan/ML_BSSN/src/Startup.ccCreating /home/jontester/Cactus/configs/sim/lib/libthorn_ML_BSSN.amake: *** [Makefile:260: sim] Error 2
Hi,
Different parts of Cactus are by default built in parallel, so a short excerpt of the build log might not be enough. We might be able to tell you more if you would log all of the output and send us that (if you know how, please compress it, it might be quite a bit).
If one of the external libraries didn't build (not unlikely), then most often that log only refers to the log of that particular library (usually called build.log). If that is the case, that file would be of particular interest.
Frank
I had been worried that on a computer with limited capabilities, interruption of the computations was more likely to occur during the lengthy run time, and hoped to prepare for it. I then discovered that I can hibernate Ubuntu, and am thinking of using it instead of Fedora, and have also been advised that the use of checkpoints is another alternative. (In qc0-mclachlan, I note the provided code segment, but the code segment includes another line with the value 55552. Would this generally be a good value on a computer of considerably different architecture?) In any case, I would still be interested in trying to determine why Fedora does not compile Cactus properly for me. It goes pretty far before it terminates. I think I can collect the output in a text file, since inline text would be overwhelming. What would be the best procedure for submitting this information? On Monday, November 20, 2017, 11:37:11 PM PST, Frank Loeffler knarf@cct.lsu.edu wrote:
On Tue, Nov 21, 2017 at 04:33:14AM +0000, iconoklastic91@gmail.com wrote:
COMPILING arrangements/McLachlan/ML_BSSN/src/RegisterSymmetries.ccCOMPILING arrangements/McLachlan/ML_BSSN/src/Startup.ccCreating /home/jontester/Cactus/configs/sim/lib/libthorn_ML_BSSN.amake: *** [Makefile:260: sim] Error 2
Hi,
Different parts of Cactus are by default built in parallel, so a short excerpt of the build log might not be enough. We might be able to tell you more if you would log all of the output and send us that (if you know how, please compress it, it might be quite a bit).
If one of the external libraries didn't build (not unlikely), then most often that log only refers to the log of that particular library (usually called build.log). If that is the case, that file would be of particular interest.
Frank
On Thu, Nov 23, 2017 at 04:18:09AM +0000, iconoklastic91@gmail.com wrote:
before it terminates. I think I can collect the output in a text file, since inline text would be overwhelming. What would be the best procedure for submitting this information?
The text file would quite likely indeed be a bit large, but it should compress quite well. Maybe the best way would be to open a ticket at
https://trac.einsteintoolkit.org/newticket
and upload the compressed build log there.
Frank
I had OpenMPI installed as well, and it seemed to be conflicting with the mpich thing. After I removed it there were no further issues on Fedora 27, though I may continue to use Ubuntu because .deb packages are easier to find. On Wednesday, November 22, 2017, 10:37:49 PM PST, Frank Loeffler knarf@cct.lsu.edu wrote:
On Thu, Nov 23, 2017 at 04:18:09AM +0000, iconoklastic91@gmail.com wrote:
before it terminates. I think I can collect the output in a text file, since inline text would be overwhelming. What would be the best procedure for submitting this information?
The text file would quite likely indeed be a bit large, but it should compress quite well. Maybe the best way would be to open a ticket at
https://trac.einsteintoolkit.org/newticket
and upload the compressed build log there.
Frank
users@lists.einsteintoolkit.org
-
Frank Loeffler -
iconoklastic91＠gmail.com -
Roland Haas