[
https://issues.apache.org/jira/browse/HDFS-899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12800094#action_12800094
]
Kan Zhang commented on HDFS-899:
--------------------------------
Some quick comments.
1. Client should only need to deal with Token<? extends TokenIdentifier>, not
any specific type like Token<DelegationTokenIdentifier>. A client can find out
the type of a token by calling token.getKind().
2. Mirroring JobToken in MapRed, the directory for Delegation Token in HDFS
should be o/a/h/hdfs/security/token/
3. See some more comments given by Owen in HADOOP-6373.
4. IssueDate and MaxDate are set twice, first in DelegationTokenIdentifier
constructor and then in secret manager.
5. I think createToken() method in secretManger isn't necessary. Simply use the
Token constructor when creating a token and keep createPassword() method as the
way it is in my token2.patch in HADOOP-6373 is sufficient (sequenceNumber
should be added as you did).
> Delegation Token Implementation
> -------------------------------
>
> Key: HDFS-899
> URL: https://issues.apache.org/jira/browse/HDFS-899
> Project: Hadoop HDFS
> Issue Type: New Feature
> Reporter: Jitendra Nath Pandey
> Attachments: HDFS-899.1.patch
>
>
> This jira tracks implementation of delegation token and corresponding
> changes in Namenode and DFS Api to issue, renew and cancel delegation tokens.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
