Comment 16 for bug 1580728

Reviewed: https://review.openstack.org/461912
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=564958dba6c280eb8e11ae9bbc819c7da6e204bd
Submitter: Jenkins
Branch: master

commit 564958dba6c280eb8e11ae9bbc819c7da6e204bd
Author: melanie witt <email address hidden>
Date: Tue May 2 21:47:12 2017 +0000

    Use six.text_type() when logging Instance object

    We're seeing a trace in gate jobs, for example:

      UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position
      402: ordinal not in range(128)

    when attempting to log an Instance object with a unicode display name.

    This resurfaced relatively recently because of the change in devstack
    to use the new OSJournalHandler with use_journal=True which is
    suspected of causing some deadlock issues [1] unrelated to this bug.

    The problem occurs in code that logs an entire Instance object when
    the object has a field with unicode characters in it (display_name).
    When the object is sent to logging, the UnicodeDecodeError is raised
    while formatting the log record here [2]. This implies an implicit
    conversion attempt to unicode at this point.

    I found that with the Instance object, the conversion to unicode fails
    with the UnicodeDecodeError unless the encoding 'utf-8' is explicitly
    specified to six.text_type(). And when specifying an encoding to
    six.text_type(), the argument to convert must be a string, not an
    Instance object, so this does the conversion in two steps as a utility
    function:

      1. Get the string representation of the Instance with repr()
      2. Call six.text_type(instance_repr, 'utf-8') passing the encoding
         if not six.PY3

    Closes-Bug: #1580728

    [1] https://review.openstack.org/#/c/462163
    [2] https://github.com/python/cpython/blob/2e576f5/Lib/logging/__init__.py#L338

    Change-Id: I0fc3ae02cb2e401b3240faf0d8b6aa5dc52b91fc