if genomic DNA is extracted from tissues, and the goal is to find variants such as SNP, indels, CNVs-which is the preferred method- whole exome sequencing or whole genome sequencing and why?
SNPs and short indels can be called both using WES and WGS. However, you can only call variants if they're sequenced, which with WES is limited to the exome, obviously. If you are interested in non-coding variants you need WGS. CNVs and other SVs are hard with short-read WGS and nearly impossible with WES.
But of course there is also a difference in price, so your budget will play a role as well.
Under what cases can we use RNA sequencing to find variants?
You should never design an RNA-seq experiment with the aim to find variants. However, if RNA-seq is what you have you can look at variants, but take into account that it's far from optimal and won't give you the best results.