6:53 a.m.
Not before should be reject tokens before that time, so changing to the
following as suggested would be the correct behaviour:
this.token.getIssuedAt() >= deployment.getNotBefore
Further, all adapters should allow a configurable allowed time sync here.
We should introduce a new property allowedTimeSkew with default set to 1
second. The above would then be changed to:
this.token.getIssuedAt() >= deployment.getNotBefore() -
deployment.getAllowedTimeSkew()
On Tue, 9 Apr 2019 at 13:50, Marek Posolda <mposolda(a)redhat.com> wrote:
My vote is to go with (2). It may turn to be more work as it needs
to
check all the adapters (node.js, maybe also gatekeeper etc).
But strictly said, if not-before is set for example to 10:0000, then my
understanding of not-before semantics is to check that: token is valid
if it was issued not before 10:00:00 .
Which translates to: token is valid if it was wasn't issued before
10:00:00 .
In other words, if not-before is 10:00:00 then token issued at 9:59:59
shouldn't be valid, but token issued at 10:00:00 should be valid IMO.
Marek
On 09/04/2019 09:21, Michal Hajas wrote:
> Hi,
>
> I found out that when you do logout-all (in this step realm.notBefore
value
> is set) and subsequent login very quickly it may happen that Keycloak
> returns tokens with an issuedAt value which is the same as the value of
the
> NotBeforePolicy.
>
> Such tokens are considered invalid in adapter due to this check [1].
>
> My question is, should we prevent such state? If yes what is correct
> behavior?
>
> 1. Do not generate tokens with the same issuedAt value as NotBefore
policy.
> For example, in TokenManager [2] check NotBefore value and change
> issuedAt for all tokens to (NotBefore + 1) in case they are same.
>
> or
>
> 2. Change condition [2]:
> .... && this.token.getIssuedAt() > deployment.getNotBefore();
> to:
> .... && this.token.getIssuedAt() >= deployment.getNotBefore();
>
> The later will probably require to also check other non-java adapters
> whether such check is present or not.
>
> Best regards,
> Michal
>
> [1]
>
https://github.com/keycloak/keycloak/blob/master/adapters/oidc/adapter-co...
>
> [2]
>
https://github.com/keycloak/keycloak/blob/master/services/src/main/java/o...
> _______________________________________________
> keycloak-dev mailing list
> keycloak-dev(a)lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/keycloak-dev
_______________________________________________
keycloak-dev mailing list
keycloak-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/keycloak-dev