problem with R2 reads from Miseq

[m_elena_bioinfo]

Dear, I have a problem in Miseq data. I have paired 2x150bp reads. I have identified some false positive (repetitive in different samples) due to short mapped reads. In fastq files (both R1 and R2) I have the reads complete, but sometimes only little fragments (about 35bp) of reads from R2.fastq map to genome, as PCR duplicates .
The related reads in always in *R2.fastq, and if I look into bam file this is "Second in pair".

Anyone has any idea? Is it an experimental problem or a bioinformatic one?
Thanks

[mastal]

problem with R2 reads from Miseq

What software was used to align the reads, and were the reads trimmed before alignment?

[m_elena_bioinfo]

I use bwa, samtools rmdup to remove duplicates (not -S option), gatk for local realignment and recalibration.
The reads before alignment are trimmed randomly. In the figure you can see three reads in R2 fastq file and in yellow the mapped short read. I don't know what happens to the rest of bp.