Just be careful with the mapreduce classes. I wouldn't be surprised if
we try to avoid any locally installed client.conf in MapReduce (using
only the ClientConfiguration stored inside the Job).
Will wait to hear back from you :)
Xu (Simon) Chen wrote:
Emm.. I have ~/.accumulo/config with "instance.rpc.sasl.enabled=true".
That property is indeed populated to ClientConfiguration the first time
- that's why I said the token worked initially.
Apparently, in the Hadoop portion that property is not set, as I added
some debug message to ZooKeeperInstance class. I think that's likely the
issue.
So the zookeeper instance is created in the following sequence:
https://github.com/apache/accumulo/blob/master/core/src/main/java/org/apache/accumulo/core/client/mapred/AbstractInputFormat.java#L341
https://github.com/apache/accumulo/blob/master/core/src/main/java/org/apache/accumulo/core/client/mapreduce/lib/impl/InputConfigurator.java#L671
https://github.com/apache/accumulo/blob/master/core/src/main/java/org/apache/accumulo/core/client/mapreduce/lib/impl/ConfiguratorBase.java#L361
The getClientConfiguration function calls getDefaultSearchPath()
eventually, so my ~/.accumulo/config should be searched. I think we are
close to the root cause... Will update when I find out more.
Thanks!
-Simon
On Thu, Jun 11, 2015 at 11:28 PM, Josh Elser <josh.el...@gmail.com> wrote:
> Are you sure that the spark tasks have the proper
ClientConfiguration? They
> need to have instance.rpc.sasl.enabled. I believe you should be able
to set
> this via the AccumuloInputFormat
>
> You can turn up logging org.apache.accumulo.core.client=TRACE and/or
set the
> system property -Dsun.security.krb5.debug=true to get some more
information
> as to why the authentication is failing.
>
>
> Xu (Simon) Chen wrote:
>>
>> Josh,
>>
>> I am using this function:
>>
>>
>>
https://github.com/apache/accumulo/blob/master/core/src/main/java/org/apache/accumulo/core/client/mapred/AbstractInputFormat.java#L106
>>
>> If I pass in a KerberosToken, it's stuck at line 111; if I pass in a
>> delegation token, the setConnectorInfo function finishes fine.
>>
>> But when I do something like queryRDD.count, spark eventually calls
>> HadoopRDD.getPartitions, which calls the following and get stuck in
>> the last authenticate() function:
>>
>>
https://github.com/apache/accumulo/blob/master/core/src/main/java/org/apache/accumulo/core/client/mapred/AbstractInputFormat.java#L621
>>
>>
https://github.com/apache/accumulo/blob/master/core/src/main/java/org/apache/accumulo/core/client/mapred/AbstractInputFormat.java#L348
>>
>>
https://github.com/apache/accumulo/blob/master/core/src/main/java/org/apache/accumulo/core/client/ZooKeeperInstance.java#L248
>>
>>
https://github.com/apache/accumulo/blob/master/core/src/main/java/org/apache/accumulo/core/client/impl/ConnectorImpl.java#L70
>>
>> Which essentially the same place where it would be stuck with
>> KerberosToken.
>>
>> -Simon
>>
>> On Thu, Jun 11, 2015 at 9:41 PM, Josh Elser<josh.el...@gmail.com>
wrote:
>>>
>>> What are the Accumulo methods that you are calling and what is the
error
>>> you
>>> are seeing?
>>>
>>> A KerberosToken cannot be used in a MapReduce job which is why a
>>> DelegationToken is automatically retrieved. You should still be able to
>>> provide your own DelegationToken -- if that doesn't work, that's a bug.
>>>
>>> Xu (Simon) Chen wrote:
>>>>
>>>> I actually added a flag such that I can pass in either a KerberosToken
>>>> or a DelegationTokenImpl to accumulo.
>>>>
>>>> Actually when a KerberosToken is passed in, accumulo converts it to a
>>>> DelegationToken - the conversion is where I am having trouble. I tried
>>>> passing in a delegation token directly to bypass the conversion, but a
>>>> similar problem happens, that I am stuck at authenticate on the client
>>>> side, and server side outputs the same output...
>>>>
>>>> On Thursday, June 11, 2015, Josh Elser<josh.el...@gmail.com
>>>> <mailto:josh.el...@gmail.com>> wrote:
>>>>
>>>> Keep in mind that the authentication path for DelegationToken
>>>> (mapreduce) and KerberosToken are completely different.
>>>>
>>>> Since most mapreduce jobs have multiple mappers (or reducers), I
>>>> expect we would have run into the case that the same
>>>> DelegationToken
>>>> was used multiple times. It would still be good to narrow
down the
>>>> scope of the problem.
>>>>
>>>> Xu (Simon) Chen wrote:
>>>>
>>>> Thanks Josh...
>>>>
>>>> I tested this in scala REPL, and called
>>>> DataStoreFinder.getDataStore()
>>>> multiple times, each time it seems to be reusing the same
>>>> KerberosToken object, and it works fine each time.
>>>>
>>>> So my problem only happens when the token is used in
accumulo's
>>>> mapred
>>>> package. Weird..
>>>>
>>>> -Simon
>>>>
>>>>
>>>> On Thu, Jun 11, 2015 at 5:29 PM, Josh
>>>> Elser<josh.el...@gmail.com> wrote:
>>>>
>>>> Simon,
>>>>
>>>> Can you reproduce this in plain-jane Java code? I don't
>>>> know
>>>> enough about
>>>> spark/scala, much less what Geomesa is actually do,
to know
>>>> what the issue
>>>> is.
>>>>
>>>> Also, which token are you referring to: A
KerberosToken or
>>>> a
>>>> DelegationToken? Either of them should be usable as many
>>>> times as you'd like
>>>> (given the underlying credentials are still available for
>>>> KT
>>>> or the DT token
>>>> hasn't yet expired).
>>>>
>>>>
>>>> Xu (Simon) Chen wrote:
>>>>
>>>> Folks,
>>>>
>>>> I am working on geomesa+accumulo+spark
integration. For
>>>> some reason, I
>>>> found that the same token cannot be used to
>>>> authenticate
>>>> twice.
>>>>
>>>> The workflow is that geomesa would try to create a
>>>> hadoop rdd, during
>>>> which it tries to create an AccumuloDataStore:
>>>>
>>>>
>>>>
>>>>
https://github.com/locationtech/geomesa/blob/master/geomesa-compute/src/main/scala/org/locationtech/geomesa/compute/spark/GeoMesaSpark.scala#L81
>>>>
>>>> During this process, a ZooKeeperInstance is created:
>>>>
>>>>
>>>>
>>>>
https://github.com/locationtech/geomesa/blob/rc7_a1.7_h2.5/geomesa-core/src/main/scala/org/locationtech/geomesa/core/data/AccumuloDataStoreFactory.scala#L177
>>>> I modified geomesa such that it would use kerberos to
>>>> authenticate
>>>> here. This step works fine.
>>>>
>>>> But next, geomesa calls
>>>> ConfigurationBase.setConnectorInfo:
>>>>
>>>>
>>>>
>>>>
https://github.com/locationtech/geomesa/blob/rc7_a1.7_h2.5/geomesa-compute/src/main/scala/org/locationtech/geomesa/compute/spark/GeoMesaSpark.scala#L69
>>>>
>>>> This is using the same token and the same zookeeper
>>>> URI,
>>>> for some
>>>> reason it is stuck on spark-shell, and the
following is
>>>> outputted on
>>>> tserver side:
>>>>
>>>> 2015-06-06 18:58:19,616 [server.TThreadPoolServer]
>>>> ERROR: Error
>>>> occurred during processing of message.
>>>> java.lang.RuntimeException:
>>>> org.apache.thrift.transport.TTransportException:
>>>> java.net
<http://java.net><http://java.net>.SocketTimeoutException: Read
>>>>
>>>> timed out
>>>> at
>>>>
>>>>
>>>>
org.apache.thrift.transport.TSaslServerTransport$Factory.getTransport(TSaslServerTransport.java:219)
>>>> at
>>>>
>>>>
>>>>
org.apache.accumulo.core.rpc.UGIAssumingTransportFactory$1.run(UGIAssumingTransportFactory.java:51)
>>>> at
>>>>
>>>>
>>>>
org.apache.accumulo.core.rpc.UGIAssumingTransportFactory$1.run(UGIAssumingTransportFactory.java:48)
>>>> at
>>>> java.security.AccessController.doPrivileged(Native
>>>> Method)
>>>> at
>>>> javax.security.auth.Subject.doAs(Subject.java:356)
>>>> at
>>>>
>>>>
>>>>
org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1622)
>>>> at
>>>>
>>>>
>>>>
org.apache.accumulo.core.rpc.UGIAssumingTransportFactory.getTransport(UGIAssumingTransportFactory.java:48)
>>>> at
>>>>
>>>>
>>>>
org.apache.thrift.server.TThreadPoolServer$WorkerProcess.run(TThreadPoolServer.java:208)
>>>> at
>>>>
>>>>
>>>>
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
>>>> at
>>>>
>>>>
>>>>
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
>>>> at
>>>>
>>>>
>>>>
org.apache.accumulo.fate.util.LoggingRunnable.run(LoggingRunnable.java:35)
>>>> at java.lang.Thread.run(Thread.java:745)
>>>> Caused by:
>>>> org.apache.thrift.transport.TTransportException:
>>>> java.net
<http://java.net><http://java.net>.SocketTimeoutException: Read
>>>> timed out
>>>> at
>>>>
>>>>
>>>>
org.apache.thrift.transport.TIOStreamTransport.read(TIOStreamTransport.java:129)
>>>> at
>>>>
>>>> org.apache.thrift.transport.TTransport.readAll(TTransport.java:84)
>>>> at
>>>>
>>>>
>>>>
org.apache.thrift.transport.TSaslTransport.receiveSaslMessage(TSaslTransport.java:182)
>>>> at
>>>>
>>>>
>>>>
org.apache.thrift.transport.TSaslServerTransport.handleSaslStartMessage(TSaslServerTransport.java:125)
>>>> at
>>>>
>>>>
org.apache.thrift.transport.TSaslTransport.open(TSaslTransport.java:253)
>>>> at
>>>>
>>>>
>>>>
org.apache.thrift.transport.TSaslServerTransport.open(TSaslServerTransport.java:41)
>>>> at
>>>>
>>>>
>>>>
org.apache.thrift.transport.TSaslServerTransport$Factory.getTransport(TSaslServerTransport.java:216)
>>>> ... 11 more
>>>> Caused by: java.net <http://java.net>
>>>> <http://java.net>.SocketTimeoutException: Read timed
>>>> out
>>>> at
>>>> java.net.SocketInputStream.socketRead0(Native Method)
>>>> at
>>>>
>>>> java.net.SocketInputStream.read(SocketInputStream.java:152)
>>>> at
>>>>
>>>> java.net.SocketInputStream.read(SocketInputStream.java:122)
>>>> at
>>>>
>>>> java.io.BufferedInputStream.read1(BufferedInputStream.java:273)
>>>> at
>>>>
>>>> java.io.BufferedInputStream.read(BufferedInputStream.java:334)
>>>> at
>>>>
>>>>
>>>>
org.apache.thrift.transport.TIOStreamTransport.read(TIOStreamTransport.java:127)
>>>> ... 17 more
>>>>
>>>> Any idea why?
>>>>
>>>> Thanks.
>>>> -Simon
>>>>
>
