The short answer is, however a species survived, it did. In humans and chimps, for example, there are some 200,000 ERV and ERV fragments, and here we still are. How do we know the ERVs are of retroviral origin? See here.
Some individuals in a population will have natural resistance to any given retrovirus. They will reproduce more successfully, conferring their resistance to subsequent generations. As the next retrovirus comes along, the individuals resistant to that will add their resistance to the gene pool.
Retroviruses convert their RNA genomes into DNA by a process called reverse transcription, which is very error-prone. Viruses do not need error checking when reverse transcribing, because they reproduce on a massive scale, and they reproduce very rapidly. Indeed this is why they can evolve so rapidly. It follows that a proportion of integrations will not be successful from the point of view of the retrovirus, and will be harmless, and even, rarely, helpful to the host organism. ERVs represent copies of at least harmless integrations, otherwise they would not have entered the germ-line and been reproduced over generations. The harmful ones, of course, would have not been reproduced.
The rate at which new ERVs enter the gene-pool of a species and become fixed (present in all members of the species) is difficult to estimate. Being mostly of neutral fitness value, they are subject to mutation and genetic recombination making their structure erode over time and become difficult to identify as ERVs for sure. The oldest identified ERV at the time of writing is perhaps a bovine BoERV1 which is estimated to be 56 - 128 million years old. Let's use a figure of 100 million years just for simplicity. 200,000 endogenizations (where a retrovirus enters the genome and becomes fixed in the population) in 100,000,000 years means one endogenization every 500 years. It doesn't sound as impossible as some creationists try to make out.