[ https://issues.apache.org/jira/browse/HDFS-12359?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16148375#comment-16148375 ]
Xiao Chen commented on HDFS-12359: ---------------------------------- Thanks for the review Wei-Chiu, added a few specific checks in patch 2. Existing reencryption tests should cover this on a higher level. bq. getCurrentKeyVersion would never return null. A little confused by this comment: # {{getCurrentKeyVersion}} could return null, if edek's {{getCurrentKeyVersion}} is null. Even if in KMS we check it, it feels to me safeguarding against nullity here does more good than harm. # {{FSDirEncryptionZoneOp.getKeyNameForZone}} could also return null, since it calls to: {code:title=EncryptionZoneManager#getKeyName} String getKeyName(final INodesInPath iip) { assert dir.hasReadLock(); EncryptionZoneInt ezi = getEncryptionZoneForPath(iip); if (ezi == null) { return null; } return ezi.getKeyName(); } {code} > Re-encryption should operate with minimum KMS ACL requirements. > --------------------------------------------------------------- > > Key: HDFS-12359 > URL: https://issues.apache.org/jira/browse/HDFS-12359 > Project: Hadoop HDFS > Issue Type: Bug > Components: encryption > Affects Versions: 3.0.0-beta1 > Reporter: Xiao Chen > Assignee: Xiao Chen > Attachments: HDFS-12359.01.patch, HDFS-12359.02.patch > > > This was caught from KMS acl testing. > HDFS-10899 gets the current key versions from KMS directly, which requires > {{READ}} acls. > It also calls invalidateCache, which requires {{MANAGEMENT}} acls. > We should fix re-encryption to not require additional ACLs than original > encryption. -- This message was sent by Atlassian JIRA (v6.4.14#64029) --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org