@tullio: vboxwrapper calls the actual application(s). In our case: python, a language interpreter. It's normal behaviour.

Surely the major distributions should work as long as you have gcc installed.

gcc --version
gcc (Ubuntu 7.3.0-16ubuntu3) 7.3.0

Something to try is to reset the project. This should delete the (large-ish) miniconda executable and download it again at the next start (which takes a couple of minutes). Beware, reset stop GPU WUs too!

This is a new install of Ubuntu 18.04, not an upgrade, and I have reset it several times.
http://www.gpugrid.net/results.php?hostid=475515&offset=0&show_names=0&state=5&appid=30

I think the previous posts show when the problem started.

Could this be a python problem that myself and others observed with new installs of linux mint 18.3?

This solved the QC error problem for us:

wget http://launchpadlibrarian.net/109052632/python-support_1.0.15_all.deb

sudo dpkg -i python-support_1.0.15_all.deb

It looks very promising, but I have reached my daily limit due to the errors. Maybe someone else can try it?

There is a similar python version issue with the FAHControl app over at Folding@home. Apparently Ubuntu 18.04 has python 3 installed so you may have to install python 2. Here is the link over at the folding forum

https://foldingforum.org/viewtopic.php?f=96&t=30778


For the QC app I've found:

linux mint 17.3, the python version is 2.7.6 and it works fine.

Linux mint 18.3, the python version is 2.7.12 and the QC app errors out.

As posted above installing the python-support_1.0.15_all.deb package solves the problem

wget http://launchpadlibrarian.net/109052632/python-support_1.0.15_all.deb

sudo dpkg -i python-support_1.0.15_all.deb

You can check the python version with:

python --version
or
python -V

On my SuSE Leap 42.3 Linux python is 2.7.13. Is this the problem? By the way, I have followed an online course on python via edx/Berkeley and I don't like it. They made use of Jupyter Notebooks which is an Interactive python developed at UC Berkeley campus, whose authors received a prize by the Association for Computing Machinery,as reported in the Berkeley Online Newsletter which I receive. But why use an interpreter? CERN uses Scientific Linux which is RedHat plus some CERN libraries, and VirtualBox in its LHC@home projects, which I run on Windows 10 and SuSE Linux. I would gladly help GPUGRID CPU tasks but they all fail.
Tullio

@Tullio, Is there a way on openSuse to "force" the QC app to use python 2.6?

At the folding forum, there are instructions for openSuse users to set up fahcontrol to use python 2.6 vs 2.7.

See step 2 at this link.

https://foldingforum.org/viewtopic.php?f=89&t=30335

@Tullio, It looks like the QC app looks for python 3.6.3. I'm not sure what you need to get it working on openSuse. Sorry.

Perhaps the developers could release a version of the QC app for python 2.7?

This solved the QC error problem for us:
wget http://launchpadlibrarian.net/109052632/python-support_1.0.15_all.deb
sudo dpkg -i python-support_1.0.15_all.deb

Unfortunately, I am still getting the python error.

<core_client_version>7.9.3</core_client_version>
<![CDATA[
<message>
process exited with code 195 (0xc3, -61)</message>
<stderr_txt>
08:41:20 (3377): wrapper (7.7.26016): starting
08:41:20 (3377): wrapper (7.7.26016): starting
08:41:20 (3377): wrapper: running ../../projects/www.gpugrid.net/Miniconda3-4.3.30-Linux-x86_64.sh (-b -u -p /var/lib/boinc-client/projects/www.gpugrid.net/miniconda)
Python 3.6.3 :: Anaconda, Inc.
08:41:29 (3377): miniconda-installer exited; CPU time 7.627092
08:41:29 (3377): wrapper: running /var/lib/boinc-client/projects/www.gpugrid.net/miniconda/bin/python (pre_script.py)

CondaValueError: prefix already exists: /var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml

Traceback (most recent call last):
  File "/var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/lib/python3.6/site-packages/psi4/__init__.py", line 55, in <module>
    from . import core
ImportError: /var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/lib/python3.6/site-packages/psi4/core.so: undefined symbol: __svml_sin4

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "run-qmml.py", line 12, in <module>
    import psi4
  File "/var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/lib/python3.6/site-packages/psi4/__init__.py", line 60, in <module>
    raise ImportError("{0}".format(err))
ImportError: /var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/lib/python3.6/site-packages/psi4/core.so: undefined symbol: __svml_sin4
Traceback (most recent call last):
  File "pre_script.py", line 31, in <module>
    raise Exception("Error running qmml")
Exception: Error running qmml
08:41:39 (3377): $PROJECT_DIR/miniconda/bin/python exited; CPU time 8.040848
08:41:39 (3377): app exit status: 0x1
08:41:39 (3377): called boinc_finish(195)

I also installed Folding 7.5.0 (before installing the above python), but did not have any python problems there. Something strange is going on, but thanks for your help.

In Jim's case the key is this:

... __svml_sin4

This may be a different failure from other people's.

There is something odd in the acceleration libraries, as if they are not found. Not easy to debug. I guess you already tried the reset, right?

I guess you already tried the reset, right?

Yes, several times. And it should be a standard installation of Ubuntu 18.04, though I was getting it earlier on my Ubuntu 16.04 machine as well.

EDIT: Just to check, I reattached my Ubuntu 16.04 machine (i7-3770), and downloaded only a single work unit (running on 4 cores). But I still got a python error. Whether it is the same one as on the 18.04 machine I don't know.

<core_client_version>7.8.3</core_client_version>
<![CDATA[
<message>
process exited with code 195 (0xc3, -61)</message>
<stderr_txt>
11:19:58 (19485): wrapper (7.7.26016): starting
11:19:58 (19485): wrapper (7.7.26016): starting
11:19:58 (19485): wrapper: running ../../projects/www.gpugrid.net/Miniconda3-4.3.30-Linux-x86_64.sh (-b -u -p /var/lib/boinc-client/projects/www.gpugrid.net/miniconda)
Python 3.6.3 :: Anaconda, Inc.
11:20:05 (19485): miniconda-installer exited; CPU time 6.042732
11:20:05 (19485): wrapper: running /var/lib/boinc-client/projects/www.gpugrid.net/miniconda/bin/python (pre_script.py)
Traceback (most recent call last):
  File "/var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/lib/python3.6/site-packages/psi4/__init__.py", line 55, in <module>
    from . import core
ImportError: /var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/lib/python3.6/site-packages/psi4/core.so: undefined symbol: __svml_sin4

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "run-qmml.py", line 12, in <module>
    import psi4
  File "/var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/lib/python3.6/site-packages/psi4/__init__.py", line 60, in <module>
    raise ImportError("{0}".format(err))
ImportError: /var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/lib/python3.6/site-packages/psi4/core.so: undefined symbol: __svml_sin4
Traceback (most recent call last):
  File "pre_script.py", line 31, in <module>
    raise Exception("Error running qmml")
Exception: Error running qmml
11:23:12 (19485): $PROJECT_DIR/miniconda/bin/python exited; CPU time 76.348243
11:23:12 (19485): app exit status: 0x1
11:23:12 (19485): called boinc_finish(195)

http://www.gpugrid.net/result.php?resultid=17462427

Thanks, looks like the same error. Will need to debug it locally.

I think on some projects (maybe Rosetta or Folding, I don't recall) they are talking about recent changes concerning static versus dynamic linking of Linux/Ubuntu libraries that are causing problems, but I know nothing about that. Good luck.

When I update the kernel modules of VirtualBox after a kernel upgrade on my SuSE Learn 42.3, the installer says it found Python and proceeds.I must confess I don't know much about Python despite taking a course and I would not like to download a version that the VirtualBox installer cannot use.Since VirtualBox is vital for me to run the CERN projects of LHC@home, I am not fiddling with it. Sorry.
Tullio

There has been a recent discussion of the differences and incompatibilities between Python 2 and Python 3 on BOINC's GitHub page - primarily directed at Python as used on the server, but it may be helpful here.

Port to Python3 #2475

Keep reading through (it's a long thread): there are some pointers to useful tools like 2to3 and python-future towards the end.

The only time when GpuGrid scientists will condescend to involve themselves in dialogue with you is when they have something they want so enjoy the experience.

---

Radio Caroline, the world's most famous offshore pirate radio station.
Great music since April 1964. Support Radio Caroline Team -
Radio Caroline

__svml_sin4

I have an older install on disk so I could take a look. The old install worked so lets first check where the symbol was defined.

$ cd boinc/projects/www.gpugrid.net/miniconda/envs/qmml/
$ grep -r svml_sin4 lib/

Binary file lib/libgdma.so matches
Binary file lib/python3.6/site-packages/psi4/core.so matches

A quick peek to core.so to verify the search result.

$ objdump -t lib/python3.6/site-packages/psi4/core.so | grep svml_sin4

0000000000000000       F *UND*  0000000000000000              __svml_sin4

Undefined there as it should be. Next libgdma.so.

$ objdump -t lib/libgdma.so | grep svml_sin4
00000000004140e0 l     O .data  0000000000000008              __svml_sin4_chosen_core_func
000000000040a4c0 l     O .data.rel.ro   0000000000000060              __svml_sin4_dispatch_table
0000000000185c90 g     F .text  00000000000008b0              __svml_sin4_e9
0000000000186540 g     F .text  00000000000005f0              __svml_sin4_l9
00000000001849e0 g     F .text  0000000000000010              __svml_sin4
0000000000184a00 g     F .text  0000000000000130              __svml_sin4_chosen_core_func_init
00000000001849f0 g     F .text  0000000000000010              __svml_sin4_chosen_core_func_get_ptr

Ok. __svml_sin4 is a function and was defined in libgdma.so.

To see how things are now I made a new miniconda installation and made a new environment there according to the pre_script.py. Running Python manually and trying to import psi4 fails with the undefined symbol error. Lets see what's going on.

$ grep -r svml_sin4 lib/

Binary file lib/python3.6/site-packages/psi4/core.so matches

$ objdump -t lib/python3.6/site-packages/psi4/core.so | grep svml_sin4

0000000000000000       F *UND*  0000000000000000              __svml_sin4

Ok. The symbol is still undefined in core.so but there is no lib where it is defined. Package lists tell what happened.

                OLD                         NEW
attrs           17.4.0 py36_0               18.1.0 py36_0
blas                                        1 mkl
ca-certificates 2017.08.26 h1d4fec5_0       2018.03.07 0
certifi         2018.1.18 py36_0            2018.4.16 py36_0
chemps2         1.8.4 0                     1.8.4 0                      psi4/label/dev
cloog           0.18.0 0                    0.18.0 0
dftd3           3.2.0 7                     3.2.0 h0e1e685_8             psi4/label/dev
dkh             1.2 1                       1.2 h173d85e_2               psi4/label/dev
erd             3.0.6 1                     3.0.6 1                      psi4/label/dev
gcc-5           5.2.0 1                     5.2.0 1                      psi4
gcc-5-mp        5.2.0 0                     5.2.0 0                      psi4
gdma            2.2.6 3                     2.2.6 hf4d0741_5             psi4/label/dev
gmp             6.1.2 h6c8ec71_1            6.1.2 h6c8ec71_1
h5py            2.7.0 np113py36_0           2.7.0 np113py36_0
hdf5            1.8.17 2                    1.8.17 2
intel-openmp    2018.0.0 hc7b2577_8         2018.0.0 8
isl             0.12.2 0                    0.12.2 0
libedit         3.1 heed3624_0              3.1.20170329 h6b74fdf_2
libefp          1.5b1 0                     1.5.0 h66000e6_0             psi4/label/dev
libffi          3.2.1 hd88cf55_4            3.2.1 hd88cf55_4
libgcc-ng       7.2.0 h7cc24e2_2            7.2.0 hdf63c60_3
libgfortran-ng  7.2.0 h9f7466a_2            7.2.0 hdf63c60_3
libint          1.2.1 1                     1.2.1 h73b9bb0_3             psi4/label/dev
libstdcxx-ng    7.2.0 h7a57d05_2            7.2.0 hdf63c60_3
libxc           3.0.0 6                     3.0.0 6                      psi4/label/dev
mkl             2018.0.1 h19d6760_4         2018.0.2 1
mkl_fft                                     1.0.1 py36h3010b51_0
mkl_random                                  1.0.1 py36h629b387_0
more-itertools                              4.1.0 py36_0
mpc             1.0.1 0                     1.0.1 0
mpfr            3.1.2 0                     3.1.2 0
ncurses         6 h9df7e31_2                6.1 hf484d3e_0
numpy           1.13.3 py36h3dfced4_2       1.13.3 py36hdbf6ddf_4
numpy-base                                  1.14.3 py36h9be14a7_1
openssl	1.0.2n  hb7f436b_0                  1.0.2o h20670df_0
pcmsolver       1.1.10 py36_1               1.1.10 py36_1                psi4/label/dev
pip             9.0.1 py36h6c6f9ce_4        10.0.1 py36_0
pluggy          0.6.0 py36hb689045_0        0.6.0 py36hb689045_0
psi4            1.2a1.dev611+8fe3513 py36_0 1.2a1.dev611+8fe3513 py36_0  psi4/label/dev
psi4-lt-mp      1.2a1.dev3 py36_0           1.2a1.dev3 py36_0	         psi4/label/dev
py              1.5.2 py36h29bf505_0        1.5.3 py36_0
pytest          3.3.2 py36_0                3.5.1 py36_0
python          3.6.4 hc3d631a_1            3.6.5 hc3d631a_2
readline        7 ha6073c6_4                7 ha6073c6_4
setuptools      38.4.0 py36_0               39.1.0 py36_0
simint          0.7 0	psi4                0.7	0                        psi4
six             1.11.0 py36h372c433_1       1.11.0 py36h372c433_1
sqlite          3.22.0 h1bed415_0           3.23.1 he433501_0
tk              8.6.7 hc745277_3            8.6.7 hc745277_3
wheel           0.30.0 py36hfd4bba0_1       0.31.1 py36_0
xz              5.2.3 h55aa19d_2            5.2.4 h14c3975_4
zlib            1.2.11 ha838bed_2           1.2.11 ha838bed_2

So gdma package along with a several others have been updated and that broke the psi4 package that has not been updated.

The package can be downgraded with bin/conda install -n qmml -c psi4/label/dev -c psi4 gdma=2.2.6=3 and at least manually running Python and importing psi4 works again. I haven't tested actually running tasks yet.


@Toni

I would suggest that you pin not only Psi4 package version but also all of its main dependencies.

Also. While doing some research about this I came about some information that Psi4 switched from statically linked MKL to dynamically linked MKL because static linking resulted in weird issues such as bad math results. SVML is Intel's Short Vector Math Lib and I'm not entirely sure how it relates to MKL. But I hope you have done some pre-screening of the results you have gotten back so far and checked that they are good.

Linux mint 18.3, the python version is 2.7.12 and the QC app errors out.

As posted above installing the python-support_1.0.15_all.deb package solves the problem

wget http://launchpadlibrarian.net/109052632/python-support_1.0.15_all.deb

sudo dpkg -i python-support_1.0.15_all.deb

No way.
- Fresh install of Mint 18.3
- Update install (sudo apt-get update/upgrade)
- wget http://launch....
- sudo dpkg -i python-support...
- gcc -v: gcc version 5.4.0 20160609
- python --version: python 2.7.12

Stops at 1,098% and after few minutes crashes:

ERROR conda.core.link:_execute_actions(337): An error occurred while installing package 'psi4::gcc-5-5.2.0-1'.
LinkError: post-link script failed for package psi4::gcc-5-5.2.0-1
running your command again with `-v` will provide additional information
location of failed script: /var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/bin/.gcc-5-post-link.sh
==> script messages <==
<None>

Attempting to roll back.

LinkError: post-link script failed for package psi4::gcc-5-5.2.0-1
running your command again with `-v` will provide additional information
location of failed script: /var/lib/boinc-client/projects/www.gpugrid.net/miniconda/envs/qmml/bin/.gcc-5-post-link.sh
==> script messages <==
<None>

@Juha: thanks, sounds exactly that. I'll try to pin asap. Thanks a lot.

@everybody: this is ONLY related to the "undefined symbol" error. Other errors have other causes.

T

I installed build-essential which includes gcc on my mint 18.3 clean install.

sudo apt install build-essential