[Python] How to prevent rule execution due to missing benchmark file in Snakemake?

I'm using Snakemake version 8.23.2 and encountering unexpected behavior related to benchmark files. I want to prevent rules from being re-executed solely due to missing benchmark files.

Snakefile


from datetime import datetime
NOW = datetime.now().strftime("%Y%m%d-%H%M%S")


rule all:
input:
"test"

rule test:
output:
"test"
benchmark:
"benchmark/" + NOW + ".tsv"
shell:
"""
sleep 3
touch {output}
"""


Unexpected Behavior


When I run snakemake test, the rule is executed on the first run as expected. However, on subsequent runs, it's triggered again due to the missing benchmark file:

❯ snakemake test
[...]
[Fri Oct 25 15:22:20 2024]
localrule test:
output: test
jobid: 0
benchmark: benchmark/20241025-152220.tsv
reason: Missing output files: benchmark/20241025-152220.tsv
resources: tmpdir=/tmp
[...]

Expected Behavior


I expect Snakemake to consider the rule up-to-date if the main output file exists, regardless of the benchmark file's presence:

❯ snakemake all
Assuming unrestricted shared filesystem usage.
Building DAG of jobs...
Nothing to be done (all requested files are present and up to date).

What I've Tried


I've attempted to use the --rerun-triggers option, but it doesn't seem to resolve this issue.

Use Case


I often run snakemake $(snakemake --list-target-rules) to trigger all target rules. Some target rules may have already succeeded in previous runs, and I don't want them to re-execute just because of missing benchmark files.

Is there a way to configure Snakemake to ignore missing benchmark files when determining whether a rule needs to be re-run?

Continue reading...