I would say this is a question that might not be resolved yet.
Going back to the earliest days of DNA - a billion years ago - its hard to imagine circular DNA being the first sort to show up. Its speculated that in the transition from RNA world to DNA world, early on each gene had a separate piece of nucleic acid coding for it which would not have been circular. It's been proposed for example that RNA viruses might have been a next step to cellular life. Although I can see an old electron microscopy reference, not it does not seem as if any RNA viruses with circular RNA have ever been found. How the first DNA chromosomes came about and what they looked like seems unclear to me.
In any case, though not typical, its not unusual to find linear chromosomes in prokaryotes. I could not find reference to archaea with linear chromosomes, but in prokaryotes, the link above points to the observation that even close relatives can have linear or circular chromosomes.
A remarkable property of the 8.2-Mb chromosome of S. erythraea was that, contrary to the expectations and the earlier data (Reeves et al., 1998), it proved to be circular. The chromosomes of its close relatives, Streptomyces coelicolor and Streptomyces avermitilis, are both linear, as is the chromosome of Rhodococcus sp. strain RHA1. These chromosomes are the largest in actinobacteria (8.7, 9.0 and 7.8 Mb respectively) and comprise some of longest DNA molecules in the prokaryotic world. In contrast, the shorter chromosomes of other actinobacteria are all circular.
It seems that linear vs circular genomes are a 'lifestyle' choice or connected to selection critera we don't understand yet:
Hence, it was tempting to speculate that the linearity of streptomycetal chromosomes was somehow linked to and perhaps beneficial for the maintenance of these extremely long DNA molecules, even though it resulted in the instability of their termini. This idea has been contradicted by the discovery of equally long but circular chromosomes in the acidobacterium Solibacter usitatus (9.97 Mb), δ-proteobacterium Myxococcus xanthus (9.1 Mb), cyanobacterium Trichodesmium erythraeum (7.8 Mb), and, now, in the actinobacterium S. erythraea.
I really doubt the instability of the chromosomal terminii is a real consideration - there must be genes that help maintain those, just as there are in eukaryotes. S coelicolor is a soil bacterium that is so common most people can find it in the dirt right outside their door - its not hard to live with a linear chromosome as a bacterium.