Hey @CuriousGeneticist !
Are you using a cluster or a laptop/server? It sounds like you’ve specified something in spark_conf. Can you share exactly how you initialized Hail? If you’re on a laptop/server you should initialize Hail as specified in this post.
Exporting the genotypes as a TSV file, as you’re doing above, converts from a relatively compact VCF-like format (one row per variant) to a relatively verbose format (one row per genotype).
Do you want the genotypes or just the variant names? If you just want the variant names do this:
mt_lof.rows().select().export('./GT.tsv')
if you want the genotypes, can you use a VCF export instead? That will be faster and smaller than a TSV export. Take a look at hl.export_vcf.