Troubleshooting

Files extracted from ZIP archives have lost their sticky bit (S_ISVTX)

Due to compatibility and security reasons, ZIP member files will have the sticky bit turned off when extracted from their parent archive. You can manually set the sticky bit afterward to turn them on. Notice that the sticky bit for directories will not be affected.

IBM z/OS XL C/C++ failures

When using pip to install packages, they might require a C/C++ compiler. See Using IBM C/C++ compilers with IBM Open Enterprise SDK for Python 3.11 for more information on setting up a C/C++ compiler.

#include <stdio.h>

int main()
{
    printf("Hello, world!\n");
    return 0;
}

If successful, the program will print out Hello, world! which means your compiler is ready to use with IBM Open Enterprise SDK for Python.

EDC5129I No such file or directory. (errno2=0x05940252) FSUM3221 xlc: Cannot spawn program /usr/lpp/cbclib/xlc/exe/ccndrvr - error: command '/bin/xlc' failed with exit code 255

Import Error: CEE3512S An HFS load of module <path to filename.so> failed. The system return code is 0000000130 and the reason code is 0BDF0C27.

ls -alT <path to filename.so>

t ISO8859-1 T=on

chtag -r <filename.so>

Fatal Python Error: Failed to get random numbers

Fatal Python error: _Py_HashRandomization_Init: failed to get random numbers to initialize Python

The above error shows up when you try to run Python without enabling Integrated Cryptographic Services Facility (ICSF). ICSF is typically responsible for supplying random data for /dev/urandom. You can run the following command to verify whether ICSF is enabled or not on your system.

head -c10 /dev/urandom

If ICSF is enabled, you will see random data and if it is not enabled on your system, you will encounter Internal Error.

For more information including installation guide, please refer to ICSF System Programmer's Guide and ICSF Administrator's Guide.

Python execution failure due to semaphore exhaustion

When semaphore exhaustion occurs on a machine, it can cause Python programs to fail in various ways. One common example will be an error message as follows:

_multiprocessing.SemLock(PermissionError: [Errno 139] EDC5139I Operation not permitted.

To diagnose whether semaphore exhaustion is an issue, you can run the following command to examine the number of semaphores that your user is currently using.

ipcs | grep <your ID>

If you run ipcs -y, you can get the limits for semaphores or the shared memory. If the number reported is close to that number, then it's likely that you are hitting the limit when running Python.

The following example shows how to approach cleaning up semaphore usage under your user id:

ipcs | grep <your ID> | awk '{print $2}' > semaphores_example.txt
for i in $(cat semaphores_example.txt) ; do { ipcrm -s $i >> /dev/null 2>&1; }; done;
for i in $(cat semaphores_example.txt) ; do { ipcrm -m $i >> /dev/null 2>&1; }; done;
rm semaphores_example.txt

Not found error for encodings module

You might run a Python script and get encodings module not found errors. The presence of the PYTHONHOME environment variable can lead to the mixing of Python versions:

$ python3
Fatal Python error: initfsencoding: unable to load the file system codec
ModuleNotFoundError: No module named 'encodings'

Current thread 0x26b7980000000001 (most recent call first):
CEE5207E The signal SIGABRT was received.
ABORT instruction

$ unset PYTHONHOME
$ python3

Errors when using distutils

Extended precision floating point support issue in NumPy library

The NumPy library in IBM Open Enterprise SDK for Python does not support direct string conversion to the long double data type. Instead, literals and strings are parsed to a double precision floating point number followed by a conversion to the long double number. This indirect conversion introduces precision and range problems for numbers outside of the double precision range.

Errors for incorrectly tagged files

SyntaxError: Non-UTF-8 code starting with '\x84' in file test.py on line 1, but no encoding declared; see http://python.org/dev/peps/pep-0263/ for details

Issues building and installing PyPI packages

Not all packages can be built by using the default options provided in Python, and a given PyPI package may not contain xlc compiler definitions. You can often avoid compile failures, for example, changing C standard level, by using the environment variable CFLAGS.

For packages requiring C11 and above, see Using IBM C/C++ compilers with IBM Open Enterprise SDK for Python 3.11 for obtaining the recommended compiler.

Import Error: CEE3512S An HFS load of module <path to filename.so> failed. The system return code is 0000000130 and the reason code is 0BDF0C27.

ls -alT <path to filename.so>

t ISO8859-1 T=on

chtag -r <filename.so>

NumPy exec_command return value issue

_BPXK_AUTOCVT='ON'

Packages unable to install cffi within a virtual environment

c/_cffi_backend.c:15:10: fatal error: 'ffi.h' file not found

<path to python3 install>/bin/python3 -m venv --system-site-packages venv

Redirecting to a file in a shell script results in garbled output

export _TAG_REDIR_ERR=txt
export _TAG_REDIR_IN=txt
export _TAG_REDIR_OUT=txt

UnicodeDecodeError error message

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xa3 in position 0: invalid start byte

Running pip without AUTOCVT results in corrupted file generation

Running pip without AUTOCVT causes extra files to be created. These file names use conflicting encodings that results difficulty in deleting the files.

To resolve corrupt file generation, you need to set up the correct environment variables following the instructions in the IBM Open Enterprise SDK for Python documents. For more information on how to set up environment variables, refer toSet environment variables.

Error installing PyPI package using requirements.txt file encoded in EBCDIC

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xa4 in position 0: invalid start byte

iconv -T -f IBM-1047 -t ISO8859-1 requirements.txt > requirements_ascii.txt

pip freeze > requirements.txt