Or this by ambiomorph at paleohacks
"There is definitely a widespread view that gluconeogenesis necessarily raises cortisol beyond what is needed without gluconeogenesis. I think the evidence is softer than it appears.
It is undisputed that cortisol stimulates gluconeogenesis. See, e.g. The Physiology of Stress: Cortisol and the Hypothalamic-Pituitary-Adrenal Axis. This leads to the claim that continual gluconeogenesis requires continually raised levels of cortisol. That's not logically implied and I'm not sure it's true. For example, Dr. Eades says, in response to a related question:
[M]ost of the time during fasting, the hormone glucagon,which is the counter-regulatory hormone to insulin, drives gluconeogenesis, not the stress hormone cortisol.
This seems consistent with this primer on glucagon physiology, which says:
When hypoglycemia is produced in humans by injection of insulin, release of glucagon is stimulated along with that of other counterregulatory hormones when the plasma glucose decreases below 3.8 mM (~68 mg/dl) (Figure 11). Restoration of euglycemia is due to a compensatory increase in hepatic glucose production. Although secretion of catecholamines, growth hormone, and cortisol are stimulated along with that of glucagon, only the increases in plasma glucagon and catecholamines coincide with or precede the compensatory increase in the glucose production rate (66,67). That glucagon is the major acute glucose counterregulatory hormone is suggested by the fact that inhibition of the plasma glucagon responses by somatostatin markedly attenuates the compensatory increase in the glucose production rate and impairs restoration of euglycemia following insulin administration (Figure 12). Prevention of cortisol secretion (68), adrenergic blockade (66), adrenalectomy (65), or acute growth hormone deficiency (66) does not appreciably affect immediate glucose counterregulation. The effects of glucagon during restoration of euglycemia involves both glycogenolysis and gluconeogenesis, predominantly the former (69).
This does, suggest, however, that whenever blood sugar goes sufficiently low, cortisol rises. Note that this is no more likely to occur in a ketogenic state than a non-ketogenic state. In fact, most people report much more stable blood sugar levels on a ketogenic diet than otherwise. It may be relevant for fasting, though. If there is a regular stream of dietary protein available for conversion to glucose, blood sugar will probably not drop so low as to raise the cortisol alarm. This is consistent with Chris Kresser's post on [i]ntermittent fasting, cortisol and blood sugar, in which he says that some of his patients, already on a low carb, paleo diet, cannot successfully do intermittent fasting because of cortisol dysregulation. In these cases, he has them eat every 2-3 hours, and they improve. He never suggests, though, that they should abandon carbohydrate restriction.
Based on this information, I believe the notion that ketosis is stressful is simply a pervasive myth. Gluconeogenesis is mediated by glucagon, not cortisol. Cortisol will only be raised if there is hypoglycemia from fasting or from excess insulin. In fact, since the best way to control insulin is to lower carbs, ketogenic diets probably reduce the need for cortisol by providing better blood sugar control.
Why do people consider ketosis "stressful" to the body? - PaleoHacks.com