So, consider the source. Providr is not a science site, it is an Interesting News Site, with writers who are likely not that savvy about what studies do and do not mean. Providr reports on a presentation at the American Association for the Advancement of Science. I don't know if that's a legit organisation. Presumably so. I didn't look up what the actual study says, because in this context it doesn't much matter. What I wish to highlight is the choice of words, which are misleading.
The study purportedly found that people who drank one or two glasses of beer or wine per day had 18% less early death. People who exercised 15-45 minutes per day had 11% less early death. But the website didn't report it that way. Providr medical writer Brandon Marji phrases it that these activities reduce early death. Not the same thing. The people who already drink 1-2 glasses a day are different people than those who drink 0, and those who drink 4-5, right out of the gate. The people who exercise 15-45 minutes a day are different people than those who don't exercise at all or those who exercise more than and hour a day. Those behaviors may identify a certain type of person, they don't (necessarily) create that sort of person. Blondes have more hairs on their head than brunettes, but you don't get more hairs by dyeing them. Basketball players are taller than others but you don't get taller by playing basketball.
Studies can try to account for a lot of variables and see past them, controlling for age, education, sex, diet, region, race, whatever. But when one is studying behaviors that people already do, there is always the possibility that some underlying cause is being overlooked.
The effect doesn't have to be large, as they studied people for years.
Starting a moderate drinking program, or a moderate exercise program, may indeed be good for you. But there is nothing in this news report, which is where most people get their health information, that is evidence for it.