Andrew,

Thanks loads for verifying this, and giving me pointers for work around. I may 
contact you if I have any other issues, but the 'fixes' you gave me should get 
us up a running!

Doing an upgrade from Essex to Havana, integrating Neutron, and dealing w 
Hadoop, I'm never sure if it's a bug on my part or a real issue. Thanks loads.

Ross

Thanks again and regards
Ross

(finger tapped on my iPhone)

On Feb 25, 2014, at 7:41 PM, "Andrew Lazarev" 
<alaza...@mirantis.com<mailto:alaza...@mirantis.com>> wrote:

Hi Ross,

I tried your scenario and was able to reproduce the issue. So, I filed bug 
https://bugs.launchpad.net/savanna/+bug/1284906 to track that.

Root cause of the issue is that auth check is missing in getData method of 
swift driver for hadoop. So, it fails to read data if this is the first 
requested operation (driver checks if destination directory exists when 
destination is in swift, that's why swift->swift works fine).

I've created patch for hadoop-swift driver: 
https://review.openstack.org/#/c/76405/. Jar and images will be updated once it 
is reviewed by community and merged.

I see several workarounds for you:
1. Bug is in recent code that adds data locality feature support. If you don't 
plan to use data locality in initial testing you can use older version of 
hadoop-swift driver 
(http://savanna-files.mirantis.com/hadoop-swift/hadoop-swift-2013-07-08.jar). 
You should place this file on all nodes as 
/usr/share/hadoop/lib/hadoop-swift.jar.
2. Issue happens only if data locality is disabled. You can enable data 
locality in savanna. Refer to savanna docs how to do it.
3. I can send you jar with issue fixed (or you can build it by yourself from 
code with https://review.openstack.org/#/c/76405/ applied).

Thanks,
Andrew.


From: Lillie Ross-CDSR11 
<ross.lil...@motorolasolutions.com<mailto:ross.lil...@motorolasolutions.com>>
Date: Wed, Feb 26, 2014 at 2:21 AM
Subject: [Openstack] Hadoop/Swift/Savanna native filesystem integration
To: "openstack@lists.openstack.org<mailto:openstack@lists.openstack.org>" 
<openstack@lists.openstack.org<mailto:openstack@lists.openstack.org>>


All,

I’m trying to integrate and test access to/from our Swift storage ring via our 
Hadoop cluster.

After setting the HDFS permissions, I’m able to perform the following tests 
using the Swift/Hadoop file system:

distcp swift://{container}.{provider}/object 
swift://{container}.{provider}/new-object
distcp hdfs://{node}:{port}/object swift://{container}.{provider}/new-object

However, I’m not able to use distcp to move objects from Swift to HDFS storage. 
The log file from the final attempt to create the associated map task is as 
follows:

2014-02-25 22:10:38,570 DEBUG 
org.apache.hadoop.fs.swift.snative.SwiftNativeFileSystem: Initializing 
SwiftNativeFileSystem against URI
swift://data.os/blob1 and working dir swift://data.os/user/ubuntu
2014-02-25 22:10:38,593 DEBUG 
org.apache.hadoop.fs.swift.http.RestClientBindings: Filesystem 
swift://data.os/blob1 is using configurati
on keys fs.swift.service.os
2014-02-25 22:10:38,594 DEBUG org.apache.hadoop.fs.swift.http.SwiftRestClient: 
Service={os} container={data} uri={http://KEYSTONE:5
000/v2.0/tokens} tenant={test} user={tester} region={(none)} publicURL={true} 
location aware={false} partition size={4718592 KB}, buff
er size={64 KB} block size={32768 KB} connect timeout={15000}, retry count={3} 
socket timeout={60000} throttle delay={0}
2014-02-25 22:10:38,594 DEBUG 
org.apache.hadoop.fs.swift.snative.SwiftNativeFileSystem: SwiftFileSystem 
initialized
2014-02-25 22:10:38,599 DEBUG 
org.apache.hadoop.fs.swift.snative.SwiftNativeFileSystemStore: Reading 
swift://data.os/blob1 from proxy n
ode
2014-02-25 22:10:38,607 INFO org.apache.hadoop.tools.DistCp: FAIL blob1 : 
org.apache.hadoop.fs.swift.exceptions.SwiftInternalStateExcep
tion: Null Endpoint -client is not authenticated
at 
org.apache.hadoop.fs.swift.http.SwiftRestClient.checkNotNull(SwiftRestClient.java:1800)
at 
org.apache.hadoop.fs.swift.http.SwiftRestClient.pathToURI(SwiftRestClient.java:1629)
at 
org.apache.hadoop.fs.swift.http.SwiftRestClient.pathToURI(SwiftRestClient.java:1669)
at 
org.apache.hadoop.fs.swift.http.SwiftRestClient.getData(SwiftRestClient.java:711)
at 
org.apache.hadoop.fs.swift.snative.SwiftNativeFileSystemStore.getObject(SwiftNativeFileSystemStore.java:294)
at 
org.apache.hadoop.fs.swift.snative.SwiftNativeInputStream.<init>(SwiftNativeInputStream.java:103)
at 
org.apache.hadoop.fs.swift.snative.SwiftNativeFileSystem.open(SwiftNativeFileSystem.java:555)
at 
org.apache.hadoop.fs.swift.snative.SwiftNativeFileSystem.open(SwiftNativeFileSystem.java:536)
at org.apache.hadoop.fs.FileSystem.open(FileSystem.java:436)
at org.apache.hadoop.tools.DistCp$CopyFilesMapper.copy(DistCp.java:419)
at org.apache.hadoop.tools.DistCp$CopyFilesMapper.map(DistCp.java:547)
at org.apache.hadoop.tools.DistCp$CopyFilesMapper.map(DistCp.java:314)
at org.apache.hadoop.mapred.MapRunner.run(MapRunner.java:50)
at org.apache.hadoop.mapred.MapTask.runOldMapper(MapTask.java:430)
at org.apache.hadoop.mapred.MapTask.run(MapTask.java:366)
at org.apache.hadoop.mapred.Child$4.run(Child.java:255)
at java.security.AccessController.doPrivileged(Native Method)
at javax.security.auth.Subject.doAs(Subject.java:415)
at 
org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1190)
at org.apache.hadoop.mapred.Child.main(Child.java:249)

2014-02-25 22:10:47,700 INFO org.apache.hadoop.mapred.TaskLogsTruncater: 
Initializing logs' truncater with mapRetainSize=-1 and reduceR
etainSize=-1
2014-02-25 22:10:47,723 ERROR org.apache.hadoop.security.UserGroupInformation: 
PriviledgedActionException as:tester cause:java.io.IOExc
eption: Copied: 0 Skipped: 0 Failed: 1
2014-02-25 22:10:47,725 WARN org.apache.hadoop.mapred.Child: Error running child
java.io.IOException: Copied: 0 Skipped: 0 Failed: 1
at org.apache.hadoop.tools.DistCp$CopyFilesMapper.close(DistCp.java:582)
at org.apache.hadoop.mapred.MapRunner.run(MapRunner.java:57)
at org.apache.hadoop.mapred.MapTask.runOldMapper(MapTask.java:430)
at org.apache.hadoop.mapred.MapTask.run(MapTask.java:366)
at org.apache.hadoop.mapred.Child$4.run(Child.java:255)
at java.security.AccessController.doPrivileged(Native Method)
at javax.security.auth.Subject.doAs(Subject.java:415)
at 
org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1190)
at org.apache.hadoop.mapred.Child.main(Child.java:249)
2014-02-25 22:10:47,739 INFO org.apache.hadoop.mapred.Task: Runnning cleanup 
for the task
ubuntu@slave1:/var/log/hadoop/ubuntu/userlogs/job_201402252129_0006/attempt_201402252129_0006_m_000000_3$

The hadoop distcp command entered was:

master$ hadoop distcp -D fs.swift.service.os.public=true -D 
fs.swift.service.os.tenant=test -D fs.swift.service.os.username=tester -D 
fs.swift.service.os.password=supersecret swift://data.os/blob1 hdfs://<hdfs 
ip>:9000/data/blob3

Am I missing anything obvious?

Thanks and regards,
Ross



--
Ross Lillie
Distinguished Member of Technical Staff
Motorola Solutions, Inc.

motorolasolutions.com<http://motorolasolutions.com>
O: +1.847.576.0012<tel:%2B1.847.576.0012>
M: +1.847.980.2241<tel:%2B1.847.980.2241>
E: ross.lil...@motorolasolutions.com<mailto:ross.lil...@motorolasolutions.com>


<MSI-Email-Identity-sm.png>


_______________________________________________
Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Post to     : 
openstack@lists.openstack.org<mailto:openstack@lists.openstack.org>
Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack



_______________________________________________
Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Post to     : 
openstack@lists.openstack.org<mailto:openstack@lists.openstack.org>
Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack

<<inline: MSI-Email-Identity-sm.png>>

_______________________________________________
Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Post to     : openstack@lists.openstack.org
Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack

Reply via email to