Closed garethsb closed 4 years ago
We may need to exercise caution if recommending a value which has generally been determined as useful for the open internet. HSTS has the potential to cause problems if there is any desire to swap between HTTP and HTTPS for experimentation, which would require a much lower max-age.
Agreed - but then we should be careful what the test suite reports if you include a non-zero but low max-age.
Let's investigate HSTS use in HTTP libraries for automated API calls.
This is probably not something we should normatively specify in BCP-003-01, but could use an informative note.
Still have not had time to investigate this, but likely needs to be examined
Try to get this worked out for next call
Just trying to move this on a bit, given that the common ground in the quotes above is around the 1 year mark. A suggestion as a result is to include the 1 year as a RECOMMENDED in the spec document, but also include a SHOULD for being able to configure or disable this to deal with any deployment trials. The testing tool should then be updated to match and test for 1 year plus, and issue warnings for either HSTS missing altogether, or HSTS with a short max age.
Work item for Andrew to do PR based on his comment.
The spec recommends using HTTP Strict-Transport-Security (HSTS) according to RFC 6797.
Is there a guideline for what value of
max-age
to use?Netsparker says:
Qualys says:
OWASP Cheat Sheet uses an example of one year:
SCIP says:
testssl.sh used in the AMWA NMOS Testing Tool implements the HSTS test as:
But I don't know whether that guidance comes from any recognized body?