audit-2.7.5-1.el7 was built to resolve this issue.

Successfully verified manually. 

OLD (audit-2.6.5-3.el7)
=======================
# man auditctl | grep -A 5 '\-b'
       -b backlog
              Set max number of outstanding audit buffers allowed (Kernel Default=64) If all buffers are full,  the
              failure flag is consulted by the kernel for action.

       --backlog_wait_time wait_time
              Set  the  time for the kernel to wait (Kernel Default 60*HZ) when the backlog_limit is reached before
              queuing more audit events to be transferred to auditd. The number must be greater than  or  equal  to
              zero and less that 10 times the default value.

# auditctl --backlog_wait_time 0; echo $?
usage: auditctl [options]
    -a <l,a>            Append rule to end of <l>ist with <a>ction
    -A <l,a>            Add rule at beginning of <l>ist with <a>ction
    -b <backlog>        Set max number of outstanding audit buffers
                        allowed Default=64
    -c                  Continue through errors in rules
    -C f=f              Compare collected fields if available:
                        Field name, operator(=,!=), field name
    -d <l,a>            Delete rule from <l>ist with <a>ction
                        l=task,exit,user,exclude
                        a=never,always
    -D                  Delete all rules and watches
    -e [0..2]           Set enabled flag
    -f [0..2]           Set failure flag
                        0=silent 1=printk 2=panic
    -F f=v              Build rule: field name, operator(=,!=,<,>,<=,
                        >=,&,&=) value
    -h                  Help
    -i                  Ignore errors when reading rules from file
    -k <key>            Set filter key on audit rule
    -l                  List rules
    -m text             Send a user-space message
    -p [r|w|x|a]        Set permissions filter on watch
                        r=read, w=write, x=execute, a=attribute
    -q <mount,subtree>  make subtree part of mount point's dir watches
    -r <rate>           Set limit in messages/sec (0=none)
    -R <file>           read rules from file
    -s                  Report status
    -S syscall          Build rule: syscall name or number
    -t                  Trim directory watches
    -v                  Version
    -w <path>           Insert watch at <path>
    -W <path>           Remove watch at <path>
    --loginuid-immutable   Make loginuids unchangeable once set
255

# grep '\--backlog_wait_time' /etc/audit/* /usr/share/doc/audit-* -R
/usr/share/doc/audit-2.6.5/rules/10-base-config.rules:--backlog_wait_time 0

NEW (audit-2.7.6-1.el7)
=======================
# man auditctl | grep '\-b' -A 5
       -b backlog
              Set max number of outstanding audit buffers allowed (Kernel Default=64) If all buffers are full,  the
              failure flag is consulted by the kernel for action.

# auditctl --backlog_wait_time 0; echo $?
usage: auditctl [options]
    -a <l,a>            Append rule to end of <l>ist with <a>ction
    -A <l,a>            Add rule at beginning of <l>ist with <a>ction
    -b <backlog>        Set max number of outstanding audit buffers
                        allowed Default=64
    -c                  Continue through errors in rules
    -C f=f              Compare collected fields if available:
                        Field name, operator(=,!=), field name
    -d <l,a>            Delete rule from <l>ist with <a>ction
                        l=task,exit,user,exclude
                        a=never,always
    -D                  Delete all rules and watches
    -e [0..2]           Set enabled flag
    -f [0..2]           Set failure flag
                        0=silent 1=printk 2=panic
    -F f=v              Build rule: field name, operator(=,!=,<,>,<=,
                        >=,&,&=) value
    -h                  Help
    -i                  Ignore errors when reading rules from file
    -k <key>            Set filter key on audit rule
    -l                  List rules
    -m text             Send a user-space message
    -p [r|w|x|a]        Set permissions filter on watch
                        r=read, w=write, x=execute, a=attribute
    -q <mount,subtree>  make subtree part of mount point's dir watches
    -r <rate>           Set limit in messages/sec (0=none)
    -R <file>           read rules from file
    -s                  Report status
    -S syscall          Build rule: syscall name or number
    -t                  Trim directory watches
    -v                  Version
    -w <path>           Insert watch at <path>
    -W <path>           Remove watch at <path>
    --loginuid-immutable  Make loginuids unchangeable once set
    --reset-lost         Reset the lost record counter
255

# grep '\--backlog_wait_time' /etc/audit/* /usr/share/doc/audit-* -R

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHEA-2017:2008