Re: I am tired, Error, Not a host:port pair: local
> Another question I have is regarding hadoop-site.xml, where I have set > hadoop.tmp.dir to be /home/hadoop/hadoop-datastore. But hadoop is creating > dfs in /tmp/hadoop-hadoop/dfs? Are you sure that hadoop-site.xml in your config directory/modified by you is being used, not some default? micha
Re: Too many fetch failures AND Shuffle error
On 6/21/08 1:53 AM, "Sayali Kulkarni" <[EMAIL PROTECTED]> wrote: > One question still, > I currently have just 5-6 nodes. But when Hadoop is deployed on a larger > cluster, say of 1000+ nodes, is it expected that every time a new machine is > added to the cluster, you add an entry in the /etc/hosts of all the (1000+) > machines in the cluster? No. Any competent system administrator will say that the installation should be using DNS or perhaps some other distributed naming service by then. Heck, even at five nodes I would have deployed DNS. :) To get an idea of how Yahoo! runs its large installations, take a look at my presentation on the Hadoop wiki ( http://wiki.apache.org/hadoop/HadoopPresentations ).
Re: how to write a file in HDFS
On Jun 22, 2008, at 9:23 PM, Samiran Bag wrote: Can you tell me how writing to a side file in output directory can be done using c program? By using the libhdfs library, which uses jni to create a C interface to HDFS. http://wiki.apache.org/hadoop/LibHDFS -- Owen
Meet Hadoop presentation: the math from page 5
Hi, I tried to better understand slide 5 of "meet hadoop": http://wiki.apache.org/hadoop-data/attachments/HadoopPresentations/attachments/oscon-part-1.pdf The slide says is: given: –10MB/s transfer –10ms/seek –100B/entry (10B entries) –10kB/page (1B pages) updating 1% of entries (100M) takes: –1000 days with random B-Tree updates –100 days with batched B-Tree updates –1 day with sort & merge I wonder how exactly to calculate the 1000 days and 100 days. time for seeking = 100 000 000 * lg(1 000 000 000) * 10 ms = (346.034177 days) time to read all pages = 100 000 000 * lg(1 000 000 000) * (10kB/10MB/ s) = 33.7924001 days Since we might need to write all pages again we can add another 33 days, though the result is not a 1000 days, so I do something fundamentally wrong. :o Thanks for any help... Stefan
Working with XML / XQuery in hadoop
Hi Just wondering if anyone out there works with and manipulates and stores XML data using Hadoop? I've seen some threads about XML RecordReaders and people who use that XML StreamXmlRecordReader to do splits. But, has anyone implemented a query framework that will use the hadoop layer to query against the XML in their map/reduce jobs? I want to know if anyone has done an XQuery or XPath executed within a haoop job to find something within the XML stored in hadoop? I can't find any samples or anyone else out there who uses XML data vs. traditional log text data. Are there any use cases of using hadoop to work with XML and then do queries against XML in a distributed manner using hadoop? Thanks.
trouble setting up hadoop
I apologize for the severe basicness of this error, but I am in the process of getting hadoop set up. I have been following the instructions in the Hadoop quickstart. I have confirmed that bin/hadoop will give me help usage information. I am now in the stage of standalone operation. I typed in: mkdir input cp conf/*.xml input bin/hadoop jar hadoop-*-examples.jar grep input output 'dfs[a-z.]+' at which point I get: Exception in thread "main" java.lang.ClassNotFoundException: java.lang.Iterable not found in gnu.gcj.runtime.SystemClassLoader{urls=[file:/home/sjm/Desktop/hado op-0.16.4/bin/../conf/,file:/home/sjm/Desktop/hadoop-0.16.4/bin/../,file:/home/s jm/Desktop/hadoop-0.16.4/bin/../hadoop-0.16.4-core.jar,file:/home/sjm/Desktop/ha doop-0.16.4/bin/../lib/commons-cli-2.0-SNAPSHOT.jar,file:/home/sjm/Desktop/hadoo p-0.16.4/bin/../lib/commons-codec-1.3.jar,file:/home/sjm/Desktop/hadoop-0.16.4/b in/../lib/commons-httpclient-3.0.1.jar,file:/home/sjm/Desktop/hadoop-0.16.4/bin/ ../lib/commons-logging-1.0.4.jar,file:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib /commons-logging-api-1.0.4.jar,file:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/j ets3t-0.5.0.jar,file:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/jetty-5.1.4.jar, file:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/junit-3.8.1.jar,file:/home/sjm/D esktop/hadoop-0.16.4/bin/../lib/kfs-0.1.jar,file:/home/sjm/Desktop/hadoop-0.16.4 /bin/../lib/log4j-1.2.13.jar,file:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/ser vlet-api.jar,file:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/xmlenc-0.52.jar,fil e:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/jetty-ext/commons-el.jar,file:/home /sjm/Desktop/hadoop-0.16.4/bin/../lib/jetty-ext/jasper-compiler.jar,file:/home/s jm/Desktop/hadoop-0.16.4/bin/../lib/jetty-ext/jasper-runtime.jar,file:/home/sjm/ Desktop/hadoop-0.16.4/bin/../lib/jetty-ext/jsp-api.jar], parent=gnu.gcj.runtime. ExtensionClassLoader{urls=[], parent=null}} at java.net.URLClassLoader.findClass (libgcj.so.7) at java.lang.ClassLoader.loadClass (libgcj.so.7) at java.lang.ClassLoader.loadClass (libgcj.so.7) at java.lang.VMClassLoader.defineClass (libgcj.so.7) at java.lang.ClassLoader.defineClass (libgcj.so.7) at java.security.SecureClassLoader.defineClass (libgcj.so.7) at java.net.URLClassLoader.findClass (libgcj.so.7) at java.lang.ClassLoader.loadClass (libgcj.so.7) at java.lang.ClassLoader.loadClass (libgcj.so.7) at org.apache.hadoop.util.RunJar.main (RunJar.java:107) I suspect the issue is path related, though I am not certain. Could someone please point me in the right direction? Much thanks, SM
Re: Working with XML / XQuery in hadoop
-BEGIN PGP SIGNED MESSAGE- Hash: RIPEMD160 Kayla, When I first started playing with Hadoop, I created an InputFormat and RecordReader that, given an XML file, created a series of key-value pairs where the XPath of the node in the document was the key and the value of the node (if it had one) was the value. At the time, it seemed like a good idea, but turned out to be horribly slow, due to the insane number of keys that were created. It also sucked to code against. It turned out to be way faster, and way easier to code, to just pass in the name of the files to be loaded and run them through your favorite parsing implementation within the Map implementation. Alternatively, if the files are small enough, you could load the XML bytes into a sequence file, and then just read them out as BytesWritable - again, into your favorite parser. (In fact, if you're dealing with XML files below the block size of HDFS, that's probably the better way to do it.) Brian Kayla Jay wrote: | Hi | | Just wondering if anyone out there works with and manipulates and | stores XML data using Hadoop? I've seen some threads about XML | RecordReaders and people who use that XML StreamXmlRecordReader to do | splits. But, has anyone implemented a query framework that will use | the hadoop layer to query against the XML in their map/reduce jobs? | | I want to know if anyone has done an XQuery or XPath executed within | a haoop job to find something within the XML stored in hadoop? | | I can't find any samples or anyone else out there who uses XML data | vs. traditional log text data. | | Are there any use cases of using hadoop to work with XML and then do | queries against XML in a distributed manner using hadoop? | | Thanks. | | | | -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (MingW32) Comment: What is this? http://pgp.ardvaark.net iD8DBQFIYAYI3YdPnMKx1eMRA1d+AKDNfYB/oR42NONht2BT4zuHZP0SXQCgjoA7 3G0oVCxBw9Fij1nWvV58zoo= =5Eyk -END PGP SIGNATURE-
Re: Working with XML / XQuery in hadoop
Yep, we do. We have a xml Writable that uses XUM behind the scene. This has a getDom and getNode(xquery) method. In readIn we read the byte array and create the xum dom object from the byte array. Write simply triggers the BinaryCodec.serialize and we write the bytes out. However the same would work if you de/serialize xml as text, though we found that is slower than xum, though works pretty stable, since xum has other issues (you need to use BinaryCodex as jvm sigelton etc). However in general this works pretty well. Stefan On Jun 23, 2008, at 9:38 PM, Kayla Jay wrote: Hi Just wondering if anyone out there works with and manipulates and stores XML data using Hadoop? I've seen some threads about XML RecordReaders and people who use that XML StreamXmlRecordReader to do splits. But, has anyone implemented a query framework that will use the hadoop layer to query against the XML in their map/reduce jobs? I want to know if anyone has done an XQuery or XPath executed within a haoop job to find something within the XML stored in hadoop? I can't find any samples or anyone else out there who uses XML data vs. traditional log text data. Are there any use cases of using hadoop to work with XML and then do queries against XML in a distributed manner using hadoop? Thanks. ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com
login error while running hadoop on MacOSX 10.5.*
I recently installed hadoop 0.17.0 in pseudo-distributed mode on a MacOSX 10.5.3 system with software managed by Fink installed in /sw. I configured hadoop to use the stock Java 1.5.0_13 installation in /Library/Java/Home. When I attempted to run a simple map/reduce job off of the dfs after starting up the daemons, the job failed with the following Java error (501 is the ID of the user used to start the hadoop daemons and run the map/reduce job): javax.security.auth.login.LoginException: Login failed: /sw/bin/whoami: cannot find name for user ID 501 What might be causing this to occur? Manually running /sw/bin/whoami as the user in question returns the corresponding username. L.G.
Re: trouble setting up hadoop
Looks like you have not install a correct java. Make sure you have a sun java installed on your nodes and java is in your path as well JAVA_HOME should be set. I think gnu.gcj is the gnu java compiler but not a java you need to run hadoop. Check on command line this: $ java -version you should see something like this: java version "1.5.0_13" Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_13- b05-237) Java HotSpot(TM) Client VM (build 1.5.0_13-119, mixed mode, sharing) HTH On Jun 23, 2008, at 9:40 PM, Sandy wrote: I apologize for the severe basicness of this error, but I am in the process of getting hadoop set up. I have been following the instructions in the Hadoop quickstart. I have confirmed that bin/hadoop will give me help usage information. I am now in the stage of standalone operation. I typed in: mkdir input cp conf/*.xml input bin/hadoop jar hadoop-*-examples.jar grep input output 'dfs[a-z.]+' at which point I get: Exception in thread "main" java.lang.ClassNotFoundException: java.lang.Iterable not found in gnu.gcj.runtime.SystemClassLoader{urls=[file:/home/sjm/Desktop/hado op-0.16.4/bin/../conf/,file:/home/sjm/Desktop/hadoop-0.16.4/ bin/../,file:/home/s jm/Desktop/hadoop-0.16.4/bin/../hadoop-0.16.4-core.jar,file:/home/ sjm/Desktop/ha doop-0.16.4/bin/../lib/commons-cli-2.0-SNAPSHOT.jar,file:/home/sjm/ Desktop/hadoo p-0.16.4/bin/../lib/commons-codec-1.3.jar,file:/home/sjm/Desktop/ hadoop-0.16.4/b in/../lib/commons-httpclient-3.0.1.jar,file:/home/sjm/Desktop/ hadoop-0.16.4/bin/ ../lib/commons-logging-1.0.4.jar,file:/home/sjm/Desktop/ hadoop-0.16.4/bin/../lib /commons-logging-api-1.0.4.jar,file:/home/sjm/Desktop/hadoop-0.16.4/ bin/../lib/j ets3t-0.5.0.jar,file:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/ jetty-5.1.4.jar, file:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/ junit-3.8.1.jar,file:/home/sjm/D esktop/hadoop-0.16.4/bin/../lib/kfs-0.1.jar,file:/home/sjm/Desktop/ hadoop-0.16.4 /bin/../lib/log4j-1.2.13.jar,file:/home/sjm/Desktop/hadoop-0.16.4/ bin/../lib/ser vlet-api.jar,file:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/ xmlenc-0.52.jar,fil e:/home/sjm/Desktop/hadoop-0.16.4/bin/../lib/jetty-ext/commons- el.jar,file:/home /sjm/Desktop/hadoop-0.16.4/bin/../lib/jetty-ext/jasper- compiler.jar,file:/home/s jm/Desktop/hadoop-0.16.4/bin/../lib/jetty-ext/jasper- runtime.jar,file:/home/sjm/ Desktop/hadoop-0.16.4/bin/../lib/jetty-ext/jsp-api.jar], parent=gnu.gcj.runtime. ExtensionClassLoader{urls=[], parent=null}} at java.net.URLClassLoader.findClass (libgcj.so.7) at java.lang.ClassLoader.loadClass (libgcj.so.7) at java.lang.ClassLoader.loadClass (libgcj.so.7) at java.lang.VMClassLoader.defineClass (libgcj.so.7) at java.lang.ClassLoader.defineClass (libgcj.so.7) at java.security.SecureClassLoader.defineClass (libgcj.so.7) at java.net.URLClassLoader.findClass (libgcj.so.7) at java.lang.ClassLoader.loadClass (libgcj.so.7) at java.lang.ClassLoader.loadClass (libgcj.so.7) at org.apache.hadoop.util.RunJar.main (RunJar.java:107) I suspect the issue is path related, though I am not certain. Could someone please point me in the right direction? Much thanks, SM ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com
Re: login error while running hadoop on MacOSX 10.5.*
Which user runs the hadoop? It should be the same you trigger the job with. On Jun 23, 2008, at 10:29 PM, Lev Givon wrote: I recently installed hadoop 0.17.0 in pseudo-distributed mode on a MacOSX 10.5.3 system with software managed by Fink installed in /sw. I configured hadoop to use the stock Java 1.5.0_13 installation in /Library/Java/Home. When I attempted to run a simple map/reduce job off of the dfs after starting up the daemons, the job failed with the following Java error (501 is the ID of the user used to start the hadoop daemons and run the map/reduce job): javax.security.auth.login.LoginException: Login failed: /sw/bin/whoami: cannot find name for user ID 501 What might be causing this to occur? Manually running /sw/bin/whoami as the user in question returns the corresponding username. L.G. ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com
Re: login error while running hadoop on MacOSX 10.5.*
Both the daemons and the job were started using the same user. L.G. Received from Stefan Groschupf on Mon, Jun 23, 2008 at 04:34:54PM EDT: > Which user runs the hadoop? It should be the same you trigger the job with. > > On Jun 23, 2008, at 10:29 PM, Lev Givon wrote: > >> I recently installed hadoop 0.17.0 in pseudo-distributed mode on a >> MacOSX 10.5.3 system with software managed by Fink installed in /sw. I >> configured hadoop to use the stock Java 1.5.0_13 installation in >> /Library/Java/Home. When I attempted to run a simple map/reduce job >> off of the dfs after starting up the daemons, the job failed with the >> following Java error (501 is the ID of the user used to start the >> hadoop daemons and run the map/reduce job): >> >> javax.security.auth.login.LoginException: Login failed: >> /sw/bin/whoami: cannot find name for user ID 501 >> >> What might be causing this to occur? Manually running /sw/bin/whoami >> as the user in question returns the corresponding username. >> >> L.G. >> >> > > ~~~ > 101tec Inc. > Menlo Park, California, USA > http://www.101tec.com > >
Re: login error while running hadoop on MacOSX 10.5.*
The fink part and /sw confuses me. When I do a which on my os x I get: $ which whoami /usr/bin/whoami Are you using the same whoami on your console as hadoop? On Jun 23, 2008, at 10:37 PM, Lev Givon wrote: Both the daemons and the job were started using the same user. L.G. Received from Stefan Groschupf on Mon, Jun 23, 2008 at 04:34:54PM EDT: Which user runs the hadoop? It should be the same you trigger the job with. On Jun 23, 2008, at 10:29 PM, Lev Givon wrote: I recently installed hadoop 0.17.0 in pseudo-distributed mode on a MacOSX 10.5.3 system with software managed by Fink installed in / sw. I configured hadoop to use the stock Java 1.5.0_13 installation in /Library/Java/Home. When I attempted to run a simple map/reduce job off of the dfs after starting up the daemons, the job failed with the following Java error (501 is the ID of the user used to start the hadoop daemons and run the map/reduce job): javax.security.auth.login.LoginException: Login failed: /sw/bin/whoami: cannot find name for user ID 501 What might be causing this to occur? Manually running /sw/bin/whoami as the user in question returns the corresponding username. L.G. ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com
Re: login error while running hadoop on MacOSX 10.5.*
Yes; I have my PATH configured to list /sw/bin before /usr/bin. Curiously, hadoop tries to invoke /sw/bin/whoami even when I set PATH to /usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin:/usr/X11R6/bin:/usr/local/bin before starting the daemons and attempting to run the job. L.G. Received from Stefan Groschupf on Mon, Jun 23, 2008 at 04:49:23PM EDT: > The fink part and /sw confuses me. When I do a which on my os x I > get: $ which whoami /usr/bin/whoami Are you using the same whoami on > your console as hadoop? > > On Jun 23, 2008, at 10:37 PM, Lev Givon wrote: > >> Both the daemons and the job were started using the same user. >> >> L.G. >> >> Received from Stefan Groschupf on Mon, Jun 23, 2008 at 04:34:54PM EDT: >>> Which user runs the hadoop? It should be the same you trigger the job >>> with. >>> >>> On Jun 23, 2008, at 10:29 PM, Lev Givon wrote: >>> I recently installed hadoop 0.17.0 in pseudo-distributed mode on a MacOSX 10.5.3 system with software managed by Fink installed in /sw. I configured hadoop to use the stock Java 1.5.0_13 installation in /Library/Java/Home. When I attempted to run a simple map/reduce job off of the dfs after starting up the daemons, the job failed with the following Java error (501 is the ID of the user used to start the hadoop daemons and run the map/reduce job): javax.security.auth.login.LoginException: Login failed: /sw/bin/whoami: cannot find name for user ID 501 What might be causing this to occur? Manually running /sw/bin/whoami as the user in question returns the corresponding username. L.G. >>> >>> ~~~ >>> 101tec Inc. >>> Menlo Park, California, USA >>> http://www.101tec.com >>> >>> >> > > ~~~ > 101tec Inc. > Menlo Park, California, USA > http://www.101tec.com > >
realtime hadoop
Hi! I am considering using Hadoop for (almost) realime data processing. I have data coming every second and I would like to use hadoop cluster to process it as fast as possible. I need to be able to maintain some guaranteed max. processing time, for example under 3 minutes. Does anybody have experience with using Hadoop in such manner? I will appreciate if you can share your experience or give me pointers to some articles or pages on the subject. Vadim
Re: realtime hadoop
Hadoop might be the wrong technology for you. Map Reduce is a batch processing mechanism. Also HDFS might be critical since to access your data you need to close the file - means you might have many small file, a situation where hdfs is not very strong (namespace is hold in memory). Hbase might be an interesting tool for you, also zookeeper if you want to do something home grown... On Jun 23, 2008, at 11:31 PM, Vadim Zaliva wrote: Hi! I am considering using Hadoop for (almost) realime data processing. I have data coming every second and I would like to use hadoop cluster to process it as fast as possible. I need to be able to maintain some guaranteed max. processing time, for example under 3 minutes. Does anybody have experience with using Hadoop in such manner? I will appreciate if you can share your experience or give me pointers to some articles or pages on the subject. Vadim ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com
Re: login error while running hadoop on MacOSX 10.5.*
Sorry I'm not a unix expert, however the problem is clearly related to whoami since this throws an error. I run hadoop in all kind of configuration super smooth on my os x boxes. Maybe rename or move /sw/whoami for a test. Also make sure you restart the os x console since changes in .bash_profile are only picked up if you "relogin" into the command line. Sorry that is all I know and could guess.. :( On Jun 23, 2008, at 10:56 PM, Lev Givon wrote: Yes; I have my PATH configured to list /sw/bin before /usr/bin. Curiously, hadoop tries to invoke /sw/bin/whoami even when I set PATH to /usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin:/usr/X11R6/ bin:/usr/local/bin before starting the daemons and attempting to run the job. L.G. Received from Stefan Groschupf on Mon, Jun 23, 2008 at 04:49:23PM EDT: The fink part and /sw confuses me. When I do a which on my os x I get: $ which whoami /usr/bin/whoami Are you using the same whoami on your console as hadoop? On Jun 23, 2008, at 10:37 PM, Lev Givon wrote: Both the daemons and the job were started using the same user. L.G. Received from Stefan Groschupf on Mon, Jun 23, 2008 at 04:34:54PM EDT: Which user runs the hadoop? It should be the same you trigger the job with. On Jun 23, 2008, at 10:29 PM, Lev Givon wrote: I recently installed hadoop 0.17.0 in pseudo-distributed mode on a MacOSX 10.5.3 system with software managed by Fink installed in / sw. I configured hadoop to use the stock Java 1.5.0_13 installation in /Library/Java/Home. When I attempted to run a simple map/reduce job off of the dfs after starting up the daemons, the job failed with the following Java error (501 is the ID of the user used to start the hadoop daemons and run the map/reduce job): javax.security.auth.login.LoginException: Login failed: /sw/bin/whoami: cannot find name for user ID 501 What might be causing this to occur? Manually running /sw/bin/ whoami as the user in question returns the corresponding username. L.G. ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com
Re: trouble setting up hadoop
Hi Stefan, I think that did it. When I type in java -version I now get: java version "1.6.0_06" Java(TM) SE Runtime Environment (build 1.6.0_06-b02) Java HotSpot(TM) Client VM (build 10.0-b22, mixed mode, sharing) And, when I run: bin/hadoop jar hadoop-*-examples.jar grep input output 'dfs[a-z]+' I get: 08/06/23 17:03:12 INFO jvm.JvmMetrics: Initializing JVM Metrics with processName=JobTracker, sessionId= 08/06/23 17:03:13 INFO mapred.FileInputFormat: Total input paths to process : 2 08/06/23 17:03:13 INFO mapred.JobClient: Running job: job_local_1 08/06/23 17:03:13 INFO mapred.MapTask: numReduceTasks: 1 08/06/23 17:03:13 INFO mapred.LocalJobRunner: file:/home/sjm/Desktop/hadoop-0.16.4/input/hadoop-site.xml:0+178 08/06/23 17:03:13 INFO mapred.TaskRunner: Task 'job_local_1_map_' done. 08/06/23 17:03:13 INFO mapred.TaskRunner: Saved output of task 'job_local_1_map_' to file:/home/sjm/Desktop/hadoop-0.16.4/grep-temp-1561747821 08/06/23 17:03:13 INFO mapred.MapTask: numReduceTasks: 1 08/06/23 17:03:13 INFO mapred.LocalJobRunner: file:/home/sjm/Desktop/hadoop-0.16.4/input/hadoop-default.xml:0+34064 08/06/23 17:03:13 INFO mapred.TaskRunner: Task 'job_local_1_map_0001' done. 08/06/23 17:03:13 INFO mapred.TaskRunner: Saved output of task 'job_local_1_map_0001' to file:/home/sjm/Desktop/hadoop-0.16.4/grep-temp-1561747821 08/06/23 17:03:13 INFO mapred.LocalJobRunner: reduce > reduce 08/06/23 17:03:13 INFO mapred.TaskRunner: Task 'reduce_ov0kiq' done. 08/06/23 17:03:13 INFO mapred.TaskRunner: Saved output of task 'reduce_ov0kiq' to file:/home/sjm/Desktop/hadoop-0.16.4/grep-temp-1561747821 08/06/23 17:03:14 INFO mapred.JobClient: Job complete: job_local_1 08/06/23 17:03:14 INFO mapred.JobClient: Counters: 9 08/06/23 17:03:14 INFO mapred.JobClient: Map-Reduce Framework 08/06/23 17:03:14 INFO mapred.JobClient: Map input records=1125 08/06/23 17:03:14 INFO mapred.JobClient: Map output records=0 08/06/23 17:03:14 INFO mapred.JobClient: Map input bytes=34242 08/06/23 17:03:14 INFO mapred.JobClient: Map output bytes=0 08/06/23 17:03:14 INFO mapred.JobClient: Combine input records=0 08/06/23 17:03:14 INFO mapred.JobClient: Combine output records=0 08/06/23 17:03:14 INFO mapred.JobClient: Reduce input groups=0 08/06/23 17:03:14 INFO mapred.JobClient: Reduce input records=0 08/06/23 17:03:14 INFO mapred.JobClient: Reduce output records=0 08/06/23 17:03:14 INFO jvm.JvmMetrics: Cannot initialize JVM Metrics with processName=JobTracker, sessionId= - already initialized 08/06/23 17:03:14 INFO mapred.FileInputFormat: Total input paths to process : 1 08/06/23 17:03:14 INFO mapred.JobClient: Running job: job_local_2 08/06/23 17:03:14 INFO mapred.MapTask: numReduceTasks: 1 08/06/23 17:03:14 INFO mapred.LocalJobRunner: file:/home/sjm/Desktop/hadoop-0.16.4/grep-temp-1561747821/part-0:0+86 08/06/23 17:03:14 INFO mapred.TaskRunner: Task 'job_local_2_map_' done. 08/06/23 17:03:14 INFO mapred.TaskRunner: Saved output of task 'job_local_2_map_' to file:/home/sjm/Desktop/hadoop-0.16.4/output 08/06/23 17:03:14 INFO mapred.LocalJobRunner: reduce > reduce 08/06/23 17:03:14 INFO mapred.TaskRunner: Task 'reduce_448bva' done. 08/06/23 17:03:14 INFO mapred.TaskRunner: Saved output of task 'reduce_448bva' to file:/home/sjm/Desktop/hadoop-0.16.4/output 08/06/23 17:03:15 INFO mapred.JobClient: Job complete: job_local_2 08/06/23 17:03:15 INFO mapred.JobClient: Counters: 9 08/06/23 17:03:15 INFO mapred.JobClient: Map-Reduce Framework 08/06/23 17:03:15 INFO mapred.JobClient: Map input records=0 08/06/23 17:03:15 INFO mapred.JobClient: Map output records=0 08/06/23 17:03:15 INFO mapred.JobClient: Map input bytes=0 08/06/23 17:03:15 INFO mapred.JobClient: Map output bytes=0 08/06/23 17:03:15 INFO mapred.JobClient: Combine input records=0 08/06/23 17:03:15 INFO mapred.JobClient: Combine output records=0 08/06/23 17:03:15 INFO mapred.JobClient: Reduce input groups=0 08/06/23 17:03:15 INFO mapred.JobClient: Reduce input records=0 08/06/23 17:03:15 INFO mapred.JobClient: Reduce output records=0 Does this all look correct? If so, thank you so much. I really appreciate all the help! -SM On Mon, Jun 23, 2008 at 4:32 PM, Stefan Groschupf <[EMAIL PROTECTED]> wrote: > Looks like you have not install a correct java. > Make sure you have a sun java installed on your nodes and java is in your > path as well JAVA_HOME should be set. > I think gnu.gcj is the gnu java compiler but not a java you need to run > hadoop. > Check on command line this: > $ java -version > you should see something like this: > java version "1.5.0_13" > Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_13-b05-237) > Java HotSpot(TM) Client VM (build 1.5.0_13-119, mixed mode, sharing) > > HTH > > > > On Jun 23, 2008, at 9:40 PM, Sandy wrote: > > I apologize for the severe basicness of this error, but I am in the >> process >> of getting hadoop set up. I have been following the instru
Re: realtime hadoop
Vadim, Depending on the nature of your data, CouchDB (http://couchdb.org) might be worth looking into. It speaks JSON natively, and has real-time map/reduce support. The 0.8.0 release is imminent (don't bother with 0.7.2), and the community is active. We're using it for something similar to what you describe, and it's working well. Chris -- Chris Anderson http://jchris.mfdz.com
Re: realtime hadoop
> Also HDFS might be critical since to access your data you need to close the file Not anymore. Since 0.16 files are readable while being written to. >> it as fast as possible. I need to be able to maintain some guaranteed >> max. processing time, for example under 3 minutes. It looks like you do not need very strict guarantees. I think you can use hdfs as a data-storage. Don't know what kind of data-processing you do, but I agree with Stefan that map-reduce is designed for batch tasks rather than for real-time processing. Stefan Groschupf wrote: Hadoop might be the wrong technology for you. Map Reduce is a batch processing mechanism. Also HDFS might be critical since to access your data you need to close the file - means you might have many small file, a situation where hdfs is not very strong (namespace is hold in memory). Hbase might be an interesting tool for you, also zookeeper if you want to do something home grown... On Jun 23, 2008, at 11:31 PM, Vadim Zaliva wrote: Hi! I am considering using Hadoop for (almost) realime data processing. I have data coming every second and I would like to use hadoop cluster to process it as fast as possible. I need to be able to maintain some guaranteed max. processing time, for example under 3 minutes. Does anybody have experience with using Hadoop in such manner? I will appreciate if you can share your experience or give me pointers to some articles or pages on the subject. Vadim ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com
Re: realtime hadoop
Interesting. we are planning on using hadoop to provide 'near' real time log analysis. we plan on having files close every 5 minutes (1 per log machine, so 80 files every 5 minutes) and then have a m/r to merge it into a single file that will get processed by other jobs later on. do you think this will namespace will explode? I wasn't thinking of clouddb.. it might be an interesting alternative once it is a bit more stable. regards Ian Stefan Groschupf wrote: Hadoop might be the wrong technology for you. Map Reduce is a batch processing mechanism. Also HDFS might be critical since to access your data you need to close the file - means you might have many small file, a situation where hdfs is not very strong (namespace is hold in memory). Hbase might be an interesting tool for you, also zookeeper if you want to do something home grown... On Jun 23, 2008, at 11:31 PM, Vadim Zaliva wrote: Hi! I am considering using Hadoop for (almost) realime data processing. I have data coming every second and I would like to use hadoop cluster to process it as fast as possible. I need to be able to maintain some guaranteed max. processing time, for example under 3 minutes. Does anybody have experience with using Hadoop in such manner? I will appreciate if you can share your experience or give me pointers to some articles or pages on the subject. Vadim ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com
Re: realtime hadoop
We use Hadoop in a similar manner, to process batches of data in real-time every few minutes. However, we do substantial amounts of processing on that data, so we use Hadoop to distribute our computation. Unless you have a significant amount of work to be done, I wouldn't recommend using Hadoop because it's not worth the overhead of launching the jobs and moving the data around. Matt On Tue, 2008-06-24 at 13:34 +1000, Ian Holsman (Lists) wrote: > Interesting. > we are planning on using hadoop to provide 'near' real time log > analysis. we plan on having files close every 5 minutes (1 per log > machine, so 80 files every 5 minutes) and then have a m/r to merge it > into a single file that will get processed by other jobs later on. > > do you think this will namespace will explode? > > I wasn't thinking of clouddb.. it might be an interesting alternative > once it is a bit more stable. > > regards > Ian > > Stefan Groschupf wrote: > > Hadoop might be the wrong technology for you. > > Map Reduce is a batch processing mechanism. Also HDFS might be critical > > since to access your data you need to close the file - means you might > > have many small file, a situation where hdfs is not very strong > > (namespace is hold in memory). > > Hbase might be an interesting tool for you, also zookeeper if you want > > to do something home grown... > > > > > > > > On Jun 23, 2008, at 11:31 PM, Vadim Zaliva wrote: > > > >> Hi! > >> > >> I am considering using Hadoop for (almost) realime data processing. I > >> have data coming every second and I would like to use hadoop cluster > >> to process > >> it as fast as possible. I need to be able to maintain some guaranteed > >> max. processing time, for example under 3 minutes. > >> > >> Does anybody have experience with using Hadoop in such manner? I will > >> appreciate if you can share your experience or give me pointers > >> to some articles or pages on the subject. > >> > >> Vadim > >> > > > > ~~~ > > 101tec Inc. > > Menlo Park, California, USA > > http://www.101tec.com > > > > >
Re: realtime hadoop
One use case I have a question about, is using Hadoop to power a web search or other query. So the full job should be done in under a second, from start to finish. You know, you have a huge datastore, and you have to run a query against that, implemented as a MR query. Is there a way to optimize that use case, where the code doesn't change, but maybe the input parameters of the job? So a MR job could reuse the java code, and even the same JVM to avoid all of the startup costs.. I bet hadoop isn't built for that yet (and enough reasons not to support it yet).. but maybe it's a usecase that shouldn't be totally ignored. And if you think about it, this is similar to what HBase is doing, at least the query execution part.. A dedicated MR daemon running ontop of the Hadoop infrastructure, so you don't incur the cost of distributing and starting fresh MR/JVM processes across the cluster.. maybe someone would want to refactor this thought process a little bit.. Matt Kent wrote: We use Hadoop in a similar manner, to process batches of data in real-time every few minutes. However, we do substantial amounts of processing on that data, so we use Hadoop to distribute our computation. Unless you have a significant amount of work to be done, I wouldn't recommend using Hadoop because it's not worth the overhead of launching the jobs and moving the data around. Matt On Tue, 2008-06-24 at 13:34 +1000, Ian Holsman (Lists) wrote: Interesting. we are planning on using hadoop to provide 'near' real time log analysis. we plan on having files close every 5 minutes (1 per log machine, so 80 files every 5 minutes) and then have a m/r to merge it into a single file that will get processed by other jobs later on. do you think this will namespace will explode? I wasn't thinking of clouddb.. it might be an interesting alternative once it is a bit more stable. regards Ian Stefan Groschupf wrote: Hadoop might be the wrong technology for you. Map Reduce is a batch processing mechanism. Also HDFS might be critical since to access your data you need to close the file - means you might have many small file, a situation where hdfs is not very strong (namespace is hold in memory). Hbase might be an interesting tool for you, also zookeeper if you want to do something home grown... On Jun 23, 2008, at 11:31 PM, Vadim Zaliva wrote: Hi! I am considering using Hadoop for (almost) realime data processing. I have data coming every second and I would like to use hadoop cluster to process it as fast as possible. I need to be able to maintain some guaranteed max. processing time, for example under 3 minutes. Does anybody have experience with using Hadoop in such manner? I will appreciate if you can share your experience or give me pointers to some articles or pages on the subject. Vadim ~~~ 101tec Inc. Menlo Park, California, USA http://www.101tec.com