Nearly everything systematic that is known about the Norse gods was written down in Christian Iceland, two centuries or more after the conversion, by men who did not worship them. That single fact should govern how the material is read.
Snorri Sturluson is the reason a coherent Norse mythology exists at all. His Edda was written as a handbook for poets who needed to understand the old allusive vocabulary, and it is organised, explanatory and immensely readable. It is also a thirteenth-century reconstruction by a Christian politician, framed at the start with a rationalising account of the gods as migrant human kings. Where Snorri is the only source for something, it is genuinely uncertain whether he recorded a tradition or tidied one.
The Poetic Edda, preserved chiefly in a single Icelandic manuscript, is older in language and less organised, and it is the closest thing to a pre-Christian voice in the corpus. Skaldic verse is older still and harder, and because its metre resists alteration it preserves divine names and kennings with unusual reliability, even when it says almost nothing about what they meant.
Outside Iceland the evidence is fragmentary but independent, and it is worth more than its bulk suggests: Adam of Bremen on the temple at Uppsala, Saxo Grammaticus in Denmark, runic inscriptions, place names, picture stones, amulets and grave goods. Where those agree with Snorri, confidence rises. Where they do not, it is Snorri who is the outlier.