[HADOOP-14636] TestKDiag failing intermittently on Jenkins/Yetus at login from keytab

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Duplicate
• Affects Version/s: 3.0.0-beta1
• Component/s: security, test

Environment

user.name = "jenkins"
java.version = "1.8.0_131"
java.security.krb5.conf = "/testptch/hadoop/hadoop-common-project/hadoop-common/target/1499472499650/krb5.conf"
kdc.resource.dir = "src/test/resources/kdc"

hadoop.kerberos.kinit.command = "kinit"
hadoop.security.authentication = "KERBEROS"
hadoop.security.authorization = "false"
hadoop.kerberos.min.seconds.before.relogin = "60"
hadoop.security.dns.interface = "(unset)"
hadoop.security.dns.nameserver = "(unset)"
hadoop.rpc.protection = "authentication"
hadoop.security.saslproperties.resolver.class = "(unset)"
hadoop.security.crypto.codec.classes = "(unset)"
hadoop.security.group.mapping = "org.apache.hadoop.security.JniBasedUnixGroupsMappingWithFallback"
hadoop.security.impersonation.provider.class = "(unset)"
dfs.data.transfer.protection = "(unset)"
dfs.data.transfer.saslproperties.resolver.class = "(unset)"
2017-07-08 00:08:20,381 WARN  security.KDiag (KDiag.java:execute(365)) - The default cluster security is insecure

• Assignee: Unassigned
• Reporter: Steve Loughran
• Created: 08/Jul/17 11:28
• Updated: 26/Oct/17 16:49
• Resolved: 29/Sep/17 20:11

Description

The test TestKDiag is failing intermittently on Yetus builds,

org.apache.hadoop.security.KerberosAuthException: Login failure for user: [e-mail] from keytab /testptch/hadoop/hadoop-common-project/hadoop-common/target/keytab javax.security.auth.login.LoginException: Unable to obtain password from user

The tests that fail are all trying to log in using a keytab just created, the JVM isn't having any of it.

Possible causes? I can think of a few to start with

1. keytab generation
2. keytab path parameter wrong
3. JVM isn't doing the login
4. some race condition
5. Host OS
6. Other environment issues (clock, network...)

There's no recent changes in the kdiag or UGI code.

The failure is intermittent, not surfacing for me (others?) locally, which which could point at: JVM, host OS, race condition, other env issues.

Issue Links

• duplicates: HADOOP-14030 PreCommit TestKDiag failure