Probably a few times, especially around 700 million years ago, the Earth seems to have come close to freezing over for a few million years at a time. Deposits of glaciers are found interbedded with marine sediments near the equator. (The Earth’s magnetic field is nearly horizontal near the equator and nearly vertical near the poles. When lava flows cool or sediments settle, the magnetization is aligned with the field and then “frozen in”. Because lava flows and sediment layers tend to be nearly horizontal, the angle between the layering and the magnetization tells the latitude when the rock formed. Near-equatorial sites of deposition for ice-related deposits have been found many times.)
We don’t think that the Earth rolled over on its side, so the planet must have been very cold. One intriguing hypothesis is that the snowball intervals represent rises in oxygen, which oxidized and thus removed chemically-reduced greenhouse gases, thus lowering the greenhouse effect (methane plus oxygen makes carbon dioxide plus water, the water rains out rapidly, and per molecule, the carbon dioxide is less effective as a greenhouse gas than the methane was, so rise of oxygen means fall of greenhouse effect—the greenhouse is still there, but just weaker!). A snowball could develop even with the rock-weathering feedback if the cooling was fast compared to 0.5 million years—a slow stabilizer can’t stop a fast cooling. Indeed, knowing what we do about the faint young sun and the slow rock-weathering feedback, we might even have predicted the occurrence of snowball-Earth events; if we make an analogy to sports, it is likely that the powerful but slow “defense” of the rock-weathering feedback would sometimes “lose” to a “fast-break” offense of climate change.
A snowball planet would have a very high albedo, and a few million years of volcanic carbon dioxide would be required to warm enough over a snowy surface to cause melting. The isotopic composition of carbon deposited during snowball events indicates that the biosphere was greatly reduced during the snowballs. (Today, plants use light carbon preferentially, so shells and the carbonate sedimentary rocks from shells end up with the heavy carbon that is left after the plants get what they want from the in-between carbon coming out of volcanoes. If the biosphere nearly stopped producing more plants, then essentially all of the carbon would be heading for carbonate rocks, probably the inorganic equivalents of shells, and so the rocks would have intermediate carbon isotopes, getting some of the light carbon that normally would go to plants, and this is observed with snowballs.)
Huge layers of odd carbonate deposited on top of the snowball layers seem to be formed from the immense amounts of carbon dioxide released during the snowball intervals. Once the snowball melted from millions of years of volcanic carbon dioxide, the warm temperatures and high carbon dioxide would have caused very rapid, extensive rock weathering, supplying immense quantities of materials to the ocean to make carbonate rocks. Thus, the snowballs show that the rock-weathering feedback works, but slowly. And, the rock-weathering feedback relies on the warming effect of carbon dioxide.
Note also that we don’t see any way that the modern Earth is heading soon for either a snowball or a Venusian runaway greenhouse, although if you look forward hundreds of millions to billions of years, a runaway Venusian greenhouse becomes likely as the sun continues to brighten. (Oddly enough, if you removed all the carbon dioxide from the air today, you probably would get a snowball. Removing the carbon dioxide would cause cooling, which would remove much of the water vapor, causing more cooling. If you removed all the water vapor, the oceans would put more up before the Earth could freeze over. So, while the water vapor contributes more of the greenhouse effect today than does carbon dioxide, the carbon dioxide is arguably the most important greenhouse gas because it controls a lot of the water vapor.) (Note also that the study of snowball-Earth events is very difficult, with only rare records of relatively short-lived but old events. The science is evolving rapidly, and some of what you read just above may be modified fairly quickly.)