On Wed, 2012-05-16 at 14:34 -0400, Stephen Gallagher wrote:
The fact that we were keeping it in memory for the full duration
of the cache timeout meant that we would never reap the benefits
of the midpoint cache refresh.
One modification. I realized that we should maintain the same minimum
time for the midpoint cache that we do elsewhere (10 seconds), so I set
that explicitly if it's lower.
Also, I realize it looks like there's a divide-by-zero issue if the
nowait percentage is set to 100, but that's protected against already in
nss_get_config().