[collectd] 5.4 data gaps in write_graphite

[ryanL]

hi again. now that i've solved the closed_wait situation, i am
noticing gaps in some of my collected graphite metrics.

polling every 60s and writing to graphite, i am noticing every so
often null values arriving on my graphite cluster. json blob below
with timestamps, 60s apart.

[
128757752.5,
1382112420
],
[
126457060.08000001,
1382112480
],
[
null,
1382112540
],
[
null,
1382112600
],
[
123042672.09,
1382112660
],
[
127440251.12,
1382112720
],

unfortunately it's tricky to capture, as it seems fairly random and
doesn't happen very often. if i can snag a good tcpdump of it i'll
certainly forward it along.

i've also got --enable-debug compiled in, but i'm not seeing anything
useful showing up.

<Plugin syslog>
  LogLevel debug
</Plugin>

ryan