IJ29942: SENDFILE SYS CALL CAN CAUSE BLOCKINGIOERROR

APAR status

• Closed as program error.

Error description

• node running kernel 4.18.0-193.19.1.el8_2 with python3.8
  hits the following error:
  
  [root@node]# python3.8 -c "import shutil;
  shutil.copyfile('test.txt','new.txt')"
  Traceback (most recent call last):
    File "<string>", line 1, in <module>
    File "/usr/lib64/python3.8/shutil.py", line 270, in
  copyfile
      _fastcopy_sendfile(fsrc, fdst)
    File "/usr/lib64/python3.8/shutil.py", line 169, in
  _fastcopy_sendfile
      raise err
    File "/usr/lib64/python3.8/shutil.py", line 149, in
  _fastcopy_sendfile
      sent = os.sendfile(outfd, infd, offset, blocksize)
  BlockingIOError: [Errno 11] Resource temporarily
  unavailable: 'test.txt' -> 'new.txt'
  

Local fix

• In Python 3.8 and beyond, setting shutil._USE_CP_SENDFILE to
  False directs
  shutil.copyfile() to avoid calling the sendfile() system
  call.  The root
  cause of the issue involves the sendfile() system call and
  how it is handled
  in Spectrum Scale for certain Linux kernel levels.  This
  workaround causes
  Python to avoid making that system call.
  
      Here is an example:
  
      [root@node]# python3.8 -c "import shutil;
      shutil._USE_CP_SENDFILE = False
      shutil.copyfile('test.txt','new.txt')"
  

Problem summary

• In a kernel >4.10 and file-sizes being a multiple of
  page sizes, a false error is returned by sendfile() once the
  read offset reaches file size.
  

Problem conclusion

• Benefits of the solution:
  No more blocking IO errors
  Work Around:
  For Python 3.8 and beyond, setting shutil._USE_CP_SENDFILE to
  False instructs
  shutil.copyfile() to avoid calls to sendfile().
  For application which calls sendfile() directly, upgrading
  Spectrum Scale to
  a level which includes the fix of this APAR.
  Problem trigger:
  Having a file size that is a multiple of the page size
  Symptom:
  Unexpected Results/Behavior
  Platforms affected: All
  Functional Area affected: All
  Customer Impact: Suggested
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

  IJ28891

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  SPEC SCALE STD

• Fixed component ID

  5737F33AP

Applicable component levels