I have count-data from RNA-seq of miRNAs. I am trying to use Deseq() in order to find the miRNAs that are DE between control group and patients, but no matter what I do, some rows do not converge. I even replaced the call to Deseq() with the following lines:

dds <- estimateSizeFactors(dds)
nc <- counts(dds, normalized=TRUE)
filter <- rowSums(nc >= 10) >= 2
dds <- dds[filter,]
dds <- estimateDispersions(dds)
dds <- nbinomWaldTest(dds, maxit=500000)

sessionInfo( )

still, I get this error message: "6 rows did not converge in beta, labelled in mcols(object)$betaConv. Use larger maxit argument with nbinomWaldTest". The run is really slow as it is, should I really choose a larger maxit?

Instead of modifying maxit, you can examine the counts for the rows that are not converging, and consider ways to filter those out. Often these have problematic distribution of counts for the NB, e.g. they contain outliers. If you think they are ignorable (do not represent DE genes) you could just set their p-values and adj p-values to NA.