docs: Clarification of --cluster-stats docs & elaborating on the situation where job ids are not passed to the status script #1221
Conversation
…a common challenge not called out in the docs. Specically, suggestions for workarounds when the cluster submission script returns more than the job id to snakemake, and snakemake passes that entire string to the cluster-status script.... sometimes the string includes characters that prevent the script from even executing. I hope this clarification makes the feature more approachable and useful :-)
iamh2o
changed the title
|
@iamh2o I like your documentation update. I wish it had been available when I was trying to understand the
I ran into this problem with Slurm. In a multi-cluster setup, Slurm returns the name of the cluster and the job ID separated by a semi-colon (when using I submitted PR #977 to specifically address this problem. By quoting the string passed to the cluster status script as a single string, this allows your script to properly parse it. I'd appreciate any feedback on my PR. |
|
Kudos, SonarCloud Quality Gate passed!
|
…ation where job ids are not passed to the status script (snakemake#1221) * clarification on how to get --cluster-status to work. And describing a common challenge not called out in the docs. Specically, suggestions for workarounds when the cluster submission script returns more than the job id to snakemake, and snakemake passes that entire string to the cluster-status script.... sometimes the string includes characters that prevent the script from even executing. I hope this clarification makes the feature more approachable and useful :-) * Update additional_features.rst Co-authored-by: Johannes Köster <johannes.koester@uni-due.de>
Description
This PR is a small clarification and elaboration on how to get --cluster-status to work .
The current docs were not clear to me about how to get this feature to work until I read them several times and had numerous false starts. So, I tried to spell out some of the information which was implied regarding how this feature works that I believe will make this more approachable to a wider audience.
Described a frequently faced challenge that must be solved for this feature to work as stated (well, must be solved by some number of users based on my hunting for answers prior to writing this PR. Specifically, enumerating a few workarounds for when the cluster submission script returns a string with more than the job id to snakemake, and snakemake then passes that entire string to the cluster-status script, causing it to best case crash the script, worst case the string contains characters which cause the shell to not even execute the script. The current docs, as written, imply snakemake has the job id and no thought need to be give to this aspect of the feature.
My intention was to clarify how this feature operated and make brief suggestions for solutions to the string, not job id, conundrum. Which, I do not consider a snakemake bug necessarily, it seems reasonable to ask users to parse this for themselves, they just should be aware it might need to be done.
QC
I added only a handful of sentences, which I proofread.
docs/) is updated to reflect the changes or this is not necessary (e.g. if the change does neither modify the language nor the behavior or functionalities of Snakemake).