Skip to content

fix: improve experience for users using multiple copies of cve-bin-tool in parallel #4777

Open
@terriko

Description

@terriko

We have a standard set of instructions for folk using cve-bin-tool in parallel jobs that recommends that people separate the database updates from the scans as follows:

https://cve-bin-tool.readthedocs.io/en/latest/how_to_guides/multiple_scans_at_once.html

But it's easy for people to miss that, and we're not perfectly handling the database so that it just magically works, as we've seen in some recent issues including #4773

I think we can do better and have a few ideas we could implement:

  1. Make it so that you can run cve-bin-tool to do an update without requiring a filename to scan. This should maybe be an option (e.g. something like cve-bin-tool --update-only) because in a lot of cases, we want people to know immediately that they need to specify a directory to scan, otherwise they'll see it start running and come back 20 minutes later and the cache will be updated but they'll have no results to look at.
  2. Handle some sort of basic auto-detection of parallel instances and direct people to the doc link above OR automatically switch any jobs to use -u never if there's already a job running and spit out a giant warning message explaining that it was done and why with a link to the docs. Could use a lock file or something for this? This will require some finesse so it won't break people's existing setups and accidentally cause things to never update.

I'm open to better ideas if anyone has any, though.

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions