内存占用非常低时，却抛出 `std::bad_alloc` 异常

[altairwei]

描述:

在使用 basevar 工具时，观察到系统内存占用非常低 (30G / 756G) 左右，但程序偶尔会抛出 std::bad_alloc 异常，提示内存分配失败。实际上在我同时并发的十几个 basevar 任务中，大部分都能跑完，但始终会有一两个出现内存分配错误。

以下是详细信息和复现步骤：

复现步骤

1. 运行以下命令：

./bin/basevar basetype -t 20 -L output/calls/all.bam.list \
    --filename-has-samplename -R data/reference/Homo_sapiens_assembly38.fasta \
    -r chr9:1-5000000 --min-af=0.001 \
    --output-vcf output/calls/chr9_1_5000000.vcf.gz \
    --output-cvg output/calls/chr9_1_5000000.cvg.tsv.gz \
    > log/20250114_075951/chr9_1_5000000.log

3. 程序运行一段时间后，抛出以下错误：

terminate called after throwing an instance of 'std::bad_alloc'
  what():  std::bad_alloc
/usr/bin/bash: line 1: 154874 Aborted                 ./bin/basevar basetype -t 20 -L output/calls/all.bam.list --filename-has-samplename -R data/reference/Homo_sapiens_assembly38.fasta -r chr9:1-5000000 --min-af=0.001 --output-vcf output/calls/chr9_1_5000000.vcf.gz --output-cvg output/calls/chr9_1_5000000.cvg.tsv.gz > log/20250114_075951/chr9_1_5000000.log

预期行为

程序应正常运行，完成数据处理任务，而不会因内存分配失败而崩溃。

实际行为

程序偶尔会抛出 std::bad_alloc 异常，尽管系统内存占用非常低。

[altairwei]

我是用 Snakemake 执行 basevar 命令的。但奇怪的时，当我单独将失败的 job 手动运行时就能成功，难道真的是因为内存分配问题吗？

[altairwei]

下面是某个失败任务的日志：

Program start on Tue Jan 14 08:39:03 2025

[INFO] BaseVar arguments:
basevar basetype -R data/reference/Homo_sapiens_assembly38.fasta \ 
   -L output/calls/all.bam.list \ 
   -q 10 \ 
   -m 0.001 \ 
   -B 200 \ 
   -t 10 \ 
   -r chr5:115000001-120000000 \ 
   --output-vcf output/calls/chr5_115000001_120000000.vcf.gz \ 
   --output-vcg output/calls/chr5_115000001_120000000.cvg.tsv.gz \ 
   --filename-has-samplename

[INFO] Finish loading arguments and we have 10 BAM/CRAM files for variants calling.

---- Calling Intervals ----
1 - chr5:115000001-120000000

[INFO] BaseVar'll load samples id from filename directly, becuase you set --filename-has-samplename.
[INFO] Tue Jan 14 08:39:03 2025. Done for loading all samples' id from alignment files, 0 seconds elapsed.

[INFO] Tue Jan 14 08:40:06 2025. Done for creating batchfile output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.1_1.bf.gz, 62 (CPU time: 62.08) seconds elapsed.
[INFO] Tue Jan 14 08:40:06 2025. Done for creating all 1 batchfiles in chr5_115000001-120000000 and start to call variants, 63 (CPU time: 62.64) seconds elapsed in total.

[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 20000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 30000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 40000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 50000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 60000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 70000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 80000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 100000 lines.
[INFO] Have been loaded 100000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 100000 lines.
[INFO] Have been loaded 100000 lines.
[INFO] Have been loaded 100000 lines.
[INFO] Have been loaded 90000 lines.
[INFO] Have been loaded 100000 lines.
[INFO] Have been loaded 100000 lines.
[INFO] Have been loaded 100000 lines.
[INFO] Have been loaded 100000 lines.
[INFO] Tue Jan 14 08:40:08 2025. Done for creating [output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.vcf.gz.2_50, output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.cvg.gz.2_50], 2 (CPU time: 14.16) seconds elapsed.
[INFO] Tue Jan 14 08:40:08 2025. Done for creating [output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.vcf.gz.1_50, output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.cvg.gz.1_50], 2 (CPU time: [INFO] Tue Jan 14 08:40:08 2025. Done for creating [output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.vcf.gz.5_50, output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.cvg.gz.5_50], 2 (CPU time: 14.17) seconds elapsed.
14.17) seconds elapsed.
[INFO] Tue Jan 14 08:40:08 2025. Done for creating [output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.vcf.gz.9_50, output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.cvg.gz.9_50], 2 (CPU time: 14.29) seconds elapsed.
[INFO] Have been loaded 100000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Have been loaded 10000 lines.
[INFO] Tue Jan 14 08:40:08 2025. Done for creating [output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.vcf.gz.7_50, output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.cvg.gz.7_50], 2 (CPU time: 14.74) seconds elapsed.
[INFO] Tue Jan 14 08:40:08 2025. Done for creating [output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.vcf.gz.10_50, output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.cvg.gz.10_50], g/output/calls/cache_chr5_115000001_120000000/chr5_115000001_120000000.chr5_115000001-120000000.cvg.gz.7_50], 2 (CPU time: 14.74) seconds elapsed.
14.78) seconds elapsed.

[altairwei]

当我并发了 482 个单线程的 basevar 时，所有任务都成功结束了。