Thanks all, I'll be testing this over the next few days. Gunnar
On Wed, Mar 9, 2016 at 10:05 PM, Zhang, Yi (Eason) <[email protected]> wrote: > Hi Gunnar, > > From your original mail, you didn't mention what Hadoop distro you would > prefer to use. > > If you'd like to install on CDH/HDP distro, I think the current installer > can work well, if you're using vanilla Hadoop, you can try to use > 'trafodion_apache_hadoop_install' script which support installing Trafodion > on Apache Hbase0.98, it may be out of date now, I plan to update it to > support Apache Hbase1.0 and 1.1 in the near future. > > Thanks, > Eason > > > -----Original Message----- > From: Liu, Ming (Ming) [mailto:[email protected]] > Sent: Thursday, March 10, 2016 08:30 > To: [email protected] > Subject: Re: Debug Build on Existing Hadoop Environment > > Hi, Gunnar and all, > > I am afraid it is not that simple at this point. > With CDH and HDP, user can tell the installer the URI for Cloudera manager > or Ambari, so the installer can invoke REST API to : > - change the HBase settings > - restart HBase > And the default installer also can correctly find out where the required > Hadoop/Hive/HBase jar and add them into CLASSPATH in the ms.env . Since for > CDH/HDP, we know where the conf file and jar file located. > > But when you install Apache Hadoop, there is no standard location of where > you put Hadoop/HBase jar and xml config file. It is allowed to untar the > Hadoop tar into anywhere, not must to /opt. Typically people may untar it > under HOME, and probably a personalized sub-folder under HOME. > And there is no URI if you not run Ambari, so how can you answer that > prompt when you run trafodion_install? > > There was an attempt to support this before, but Trafodion is fast moving, > so I don't know if that script still works. It was called > 'trafodion_apache_hadoop_install' under the installer's 'installer' folder, > Probably not run correctly now, but good to test it. > It was Eason implemented it, so he is the best person to answer questions > if you run into issues with that script. > > There is also an ongoing effort here to support vanilla Apache Hadoop from > installer point of view, Eason just made the required change to > sqenvcom.sh. As far as I know it is still not in the github code now. > > But if Gunnar you can successfully make an installation, it will be great, > since we no longer test this for long time, so I maybe all wrong here. > > Thanks, > Ming > > -----邮件原件----- > 发件人: Venkat Muthuswamy [mailto:[email protected]] > 发送时间: 2016年3月10日 7:40 > 收件人: [email protected] > 主题: RE: Debug Build on Existing Hadoop Environment > > If you run the installer, if will prompt you for the individual tar files > (Server, dcs, rest.). Make sure you specify the fully qualified path names > to the tar files. > > Venkat > > -----Original Message----- > From: Steve Varnau [mailto:[email protected]] > Sent: Wednesday, March 09, 2016 3:25 PM > To: [email protected] > Subject: RE: Debug Build on Existing Hadoop Environment > > That's all there is to installing trafodion. The installer does not help > you with the test suites. Those you'd need to un-tar and invoke according > to their individual instructions. > > --Steve > > > -----Original Message----- > > From: Gunnar Tapper [mailto:[email protected]] > > Sent: Wednesday, March 9, 2016 2:31 PM > > To: [email protected] > > Subject: Debug Build on Existing Hadoop Environment > > > > Hi, > > > > I'd like to try out installing and testing the Trafodion Build (make > > package-all) onto an existing Hadoop environment. > > > > Is it simply a matter of moving over the tar files and then run the > > Trafodion Installer or is there more to it? > > > > -- > > Thanks, > > > > Gunnar > > *If you think you can you can, if you think you can't you're right.* > -- Thanks, Gunnar *If you think you can you can, if you think you can't you're right.*
