[
https://issues.apache.org/jira/browse/NIFI-9797?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17507771#comment-17507771
]
Tamas Palfy commented on NIFI-9797:
-----------------------------------
The "fix" provided for the issue was rushed.
It actually made things worse! _Now_ the code can result in errors that
wouldn't have been possible before.
It seems the "margin" concept was not clear.
The idea is to renew the ticket _way before_ the last possible second. To leave
us some room - a margin of error so to speak.
The current implementation _adds_ 5 extra seconds when we consider the ticket
_not_ expired - even when it _has_.
In case of a busy processor this can lead to an unnecessary failure.
Opening another ticket.
> AccessToken isExpired broken
> ----------------------------
>
> Key: NIFI-9797
> URL: https://issues.apache.org/jira/browse/NIFI-9797
> Project: Apache NiFi
> Issue Type: Bug
> Components: Core Framework
> Affects Versions: 1.16.0
> Reporter: Lawrence
> Assignee: David Handermann
> Priority: Major
> Fix For: 1.16.0
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> While using a build of the current snapshot (1.16.0) so i could use the new
> StandardOAuth2AccessTokenProvider I ran into an issue with how the
> AccessToken got changed to calculate if the token is expired. This is going
> to break potentially a lot of folks using that AccessToken. It's subtracting
> 5000 seconds from the expiresIn property. I think it was assumed that
> expires_in would be in millis also, but it is not, per OAuth standards:
> [https://datatracker.ietf.org/doc/html/rfc6749#section-4.2.2]
> Here is a link to the PR that got merged where I added a comment ( to little
> to late ).
> [https://github.com/apache/nifi/pull/5319#pullrequestreview-909366668]
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
