spark-master is not a recognized option

Hello,

I am trying to run Mutect2 (GATK v4.0.10.0) the spark-runner LOCAL mode. As I explain below I am failing to correctly request the number of local cores.

This is how I run Mutect2 with the default --spark-runner LOCAL mode

```
gatk Mutect2 \
-R GRCh38_full_analysis_set_plus_decoy_hla.fa \
--tumor-sample HCC1143_tumor \
--input hcc1143_N_subset50K.bam \
--input hcc1143_T_subset50K.bam \
--output mutect2.vcf \
-- --spark-runner LOCAL
```

Then for testing I book a one CPU machine in our cluster and I see Spark Runner is trying to use its default sparkMaster value of local[4].

The following two lines from Mutect2 logs confirm me the above
14:34:43.360 INFO IntelPairHmm - Available threads: 1
14:34:43.360 INFO IntelPairHmm - Requested threads: 4


If I try to request one CPU in gatk like this:

```
gatk Mutect2 \
-R GRCh38_full_analysis_set_plus_decoy_hla.fa \
--tumor-sample HCC1143_tumor \
--input hcc1143_N_subset50K.bam \
--input hcc1143_T_subset50K.bam \
--output mutect2.vcf \
-- --spark-runner LOCAL --spark-master local[1]
```

I get the error:
A USER ERROR has occurred: spark-master is not a recognized option


Any help to correctly use --spark-master to effectively select the number of cores will be very appreciated

Thanks a lot,
Jorge

Answers

Sign In or Register to comment.