Forward-reverse read counts don't match

[Arvind Bhagwat]

This is 16S metagenomic seq data, sent to service provider for MiSeq but they did HiSeq and 1/16th lane, 300 PE.
However, now CLC Workbench is flagging at the import paired-end read stage, saying:
"The data seems to be corrupt or originates from different sources since the input files contain different numbers of reads! The previous sequences have been properly saved but it is highly likely the import data is incomplete or defect. Please double-check the sources."
There seems to uneven number of forward and reverse reads:
2279992 A1.1.fq
1924508 A1.2.fq
463012 A2.1.fq
389656 A2.2.fq
Is this normal, how serious is the flag posted? How am I to ensure that forward and reverse reads really in pairs?

[GenoMax]

Sounds like your core may have trimmed data using BaseSpace (or some other means) which left orphan reads (they did not do PE aware trimming). You should ask them to provide raw/untrimmed original data.