The issue with poor sleep is actually poor s_low wave sleep_. This is the specific subtype of overall sleep that is reduced in depression, even when the individual sleeps too much.
Several studies have investigated the question, but a very recent one best illustrates the possible relationship between sleep and insulin/glucose control.
The study took "normal" non-diabetic subjects, and selectively reduced their slow wave sleep-- the deep sleep that is often found in naps-- but left the light sleep and the REM sleep (the dreaming sleep that runs usually about 90 minutes per cycle in most people) and studied the effects on insulin release and insulin sensitivity.
Insulin sensitivity is how strongly a given amount of insulin will affect (i.e. lower) blood sugar. High sensitivity means that a given amount can reduce the glucose a lot; low sensitivity means that even in the presence of a lot of insulin, glucose doesn't go down (much.)
Type II diabetes is characterized not by insufficient insulin, but by insulin resistance. By way of analogy, your body is so used to having a lot of glucose in it that it doesn't respond to insulin as well as it did in the past; this is why it becomes more common in overweight and/or older individuals.
The study found that the selective decrease in SWS-- even when total sleep time was the same-- caused marked decreases in insulin sensitivity.
But there's a little more to it. Importantly, people who started with low amounts of SWS had the worst outcome when (what little they had) SWS was decreased.
This is what becomes relevant to people with depression. Depression, in general, reduces SWS even when total sleep time is intact. Obviously, it reduces it even more if overall sleep is reduced as well-- but even patients with hypersomnia ("atypical depression") have decreased SWS. We already know from unrelated studies that patients with depression have higher rates of diabetes-- and this may be one of the causes.
Next up are the medications, which, as a class, have little effect on total sleep (may reduce it), reduce REM sleep, but increase SWS. This is particularly important for two reasons: first, it may have the potential to mitigate the effect of clinical depression on SWS and diabetes; second, a significant number of people on antidepressants also take sleeping pills-- pills, which, like antidepressants, typically suppress REM sleep but increase SWS and total sleep.
This is clearly not to say that sleeping pills will prevent diabetes; but it does suggest that sleep normalization in the context of the treatment of depression has an effect on the long term morbidity of depression-- e.g. Diabetes.