keep database in RAM and debug

[DocAndymomo]

I have 2 questions：
1、I will run 200+ samples，but it take too much time to load reference database into RAM when processing each sample，any method to keep database in RAM every time?
2、

krakenuniq --db /mnt/DATA3/DB_3/k2_database/02-fungi_ref --threads 20 --classified-out /mnt/06-ku --only-classified-output --preload-size 50GB --report-file /mnt/06-ku/fq1.krakenuniq.report.tsv > /mnt/DATA1/TCGA/06-ku/fq.READCLASSIFICATION.tsv --paired /mnt/05-bowtie2mr-chm/test/fq_chm_remove_R1.fastq.gz /mnt/DATA1/TCGA/05-bowtie2mr-chm/test/fq_chm_remove_R2.fastq.gz
/root/miniconda3/share/krakenuniq-1.0.4-2/libexec/classify -d /mnt/DATA3/DB_3/k2_database/02-fungi_ref/database.kdb -i /mnt/DATA3/DB_3/k2_database/02-fungi_ref/database.idx -t 20 -C /mnt/DATA1/TCGA/06-ku -c -x 50GB -r /mnt/06-ku/fq.krakenuniq.report.tsv -a /mnt/DATA3/DB_3/k2_database/02-fungi_ref/taxDB -p 12
 Database /mnt/DATA3/DB_3/k2_database/02-fungi_ref/database.kdb
Loaded database with 15672402085 keys with k of 31 [val_len 4, key_len 8].
Reading taxonomy index from /mnt/DATA3/DB_3/k2_database/02-fungi_ref/taxDB. Done.
terminate called after throwing an instance of 'std::__ios_failure'
  what():  basic_ios::clear: iostream error

how to solve this err？

Thank you!