Thanks Eric and John, I'll redo my testing tonight the RC6.
Mike On Tue, Nov 8, 2011 at 11:03 AM, John W Vines <[email protected]> wrote: > > ----- Original Message ----- > | From: "Michael Wall" <[email protected]> > | To: [email protected] > | Sent: Tuesday, November 8, 2011 10:26:59 AM > | Subject: Re: accumulo 1.3.5rc4 > | I tested RC4 on my Mac, OSX 10.6.6, with hadoop 0.20.2 and zookeeper > | 3.3.3. > | Just a local install, no cluster. > | > | +1 md5 > | +1 mvn clean package && mvn assembly:single > | > | I'll redo with RC6 and get some more of the functional tests run. > | > | Couple of items > | > | 1) When I ran bin/accumulo init, I got the following. > | ----------------------------------------------- > | 08 08:05:38,584 [util.Initialize] INFO : Hadoop Filesystem is > | hdfs://localhost:9000 > | 08 08:05:38,675 [util.Initialize] INFO : Accumulo data dir is > | /accumulo > | 08 08:05:38,675 [util.Initialize] INFO : Zookeeper server is > | localhost:2181 > | Instance name : acc-test > | Enter initial password for root: ****** > | Confirm initial password for root: ****** > | 08 08:05:51,978 [util.NativeCodeLoader] WARN : Unable to load > | native-hadoop library for your platform... using builtin-java classes > | where > | applicable > | 08 08:05:52,277 [security.ZKAuthenticator] INFO : Initialized root > | user > | with username: root at the request of user !SYSTEM > | ------------------------------------------------ > | > | Changing to src/server/src/c++ and running make built them fine and > | appears > | to have put them in the correct lib/native directory. I hadn't noticed > | before, but the message says there are built in java classes for the > | map > | and mlock. I am assuming the c++ version are just more performant. > | > | 2) Accumulo doesn't run if you just export HADOOP_HOME, ZOOKEEPER_HOME > | and > | ACCUMULO_HOME in your current shell. You get the following in the logs > | in > | every .out file. You must set those variables in your .bashrc so they > | are > | persisted when the startup scripts ssh. > | --------------------------------------------- > | less logs/master_localhost.out > | You must set HADOOP_HOME > | -------------------------------------------- > | > | 3) Ran test/system/test1. I had to create the test_ingest table by > | hand, > | which didn't seem right according to TestIngest.java. Once I did, > | ingest_test.sh, verify_test.sh, ingest_test_2.sh, verify_testt_2.sh > | and > | ingest_test_3.sh seemed to run ok. However, a scan of test_ingest > | after > | ingest_test_3.sh shows no records and takes a couple of seconds to > | complete. But the status page on http://localhost:50095 shows 3.4 > | million > | records. I set the auths to L1,L2,G1,GROUP2, which is what it looks > | like > | is being used in TestIngest.java, but still nothing is returned in the > | scan. Is this right? > | > | I'll redo with RC6. > | > | Mike Wall > | > | On Mon, Nov 7, 2011 at 1:51 PM, Keith Turner <[email protected]> wrote: > | > | > On Fri, Nov 4, 2011 at 10:23 AM, Eric Newton <[email protected]> > | > wrote: > | > > I propose that we accept > | > > > http://people.apache.org/~ecn/accumulo-1.3.5-incubating-RC4-dist.tar.gz > | > > as the official Accumulo 1.3.5 release. > | > > > | > > I generated this tarball by checking out > | > > https://svn.apache.org/repos/asf/incubator/accumulo/tags/1.3rc4 > | > > and running "./src/assembly/build.sh" > | > > > | > > The GPG signature can be found at > | > > > | > > http://people.apache.org/~ecn/accumulo-1.3.5-incubating-RC4-dist.tar.gz.asc > | > > > | > > It has an MD5 sum of ff2fc0262fd0f3f7ea92eae1b8ed96e6. > | > > > | > > Please download, verify sig/sum, and install/test. > | > > > | > > This vote closes in 72 hours. > | > > > | > > | > -1 The configuration documentation page is not correct. We are > | > thinking of doing an RC5. > | > > | > Also the build process for this page is screwy. Need to fix that for > | > 1.4. > | > > > 2. We intend for those settings to be set in accumulo-env.sh rather then > the default environment settings on every box. I'm putting in a ticket now > to clarify that in the documentation. (ACCUMULO-127) > > 3. That is an old test (there is a standing ticket for those to be > revamped). That does not disregard any failures of it though, but since the > monitor is reporting results, that leads me to believe there might be some > label issue going on. If you can't figure it out, I'll take a look at it. > But L1,L2,G1,GROUP2 should be sufficient. > > John >
