Adds a config option to customize the memory allocated to GenotypeGVCFs jobs.
Default is 15GB to match the default 4 cpus (from resources.py: ncpu = mem_gb / mem_gb_per_core = 15 / 3.75 = 4)
Again, like https://github.com/populationgenomics/production-pipelines/pull/786, this option should not be necessary, unless the scatter count is too low and you need to bump the memory to get the failing jobs over the line.
Choosing the right scatter count should mean all jobs can complete with STANDARD machine type and default memory / cpu options.
Adds a config option to customize the memory allocated to GenotypeGVCFs jobs. Default is 15GB to match the default 4 cpus (from
resources.py:ncpu = mem_gb / mem_gb_per_core = 15 / 3.75 = 4)This option is necessary because even with HIGHMEM workers, some jobs in the current run are still failing with exit code 247 (container OOM). e.g. https://batch.hail.populationgenomics.org.au/batches/455517/jobs/16
Again, like https://github.com/populationgenomics/production-pipelines/pull/786, this option should not be necessary, unless the scatter count is too low and you need to bump the memory to get the failing jobs over the line. Choosing the right scatter count should mean all jobs can complete with STANDARD machine type and default memory / cpu options.