Details

    • Type: New Feature New Feature
    • Status: Open
    • Priority: Major Major
    • Resolution: Unresolved
    • Affects Version/s: None
    • Fix Version/s: None
    • Component/s: new service
    • Labels:
      None

      Description

      Now that we can easily deploy Hadoop cluster it would be nice if we could also deploy Hive.

        Issue Links

          Activity

          Hide
          Adam Lippai added a comment -

          Installing and setting up mysql and hive on one server is eligible or you want to see different approach?

          Show
          Adam Lippai added a comment - Installing and setting up mysql and hive on one server is eligible or you want to see different approach?
          Hide
          Andrei Savu added a comment -

          It should be enough for the first iteration. What do you think?

          Show
          Andrei Savu added a comment - It should be enough for the first iteration. What do you think?
          Hide
          Paul Baclace added a comment -

          I think having the mysql (Hive metadata store) and HIVE_HOME on one server is a good start, especially since the master node is usually idle in small clusters. My understanding is that the Hive metadata store is used to compile queries into Hadoop jobs, so load level is modest, and load grows by number of job requests plus number of data inserts, multiplied by "size of data" ("number of datanodes").

          If the mysql component/role is specified is a distinctive way, like hive-mysql, then the configuration reference can later be migrated to be on a separate machine.

          I could use a Hive service in Whirr next week, actually (a patch would be fine).

          Show
          Paul Baclace added a comment - I think having the mysql (Hive metadata store) and HIVE_HOME on one server is a good start, especially since the master node is usually idle in small clusters. My understanding is that the Hive metadata store is used to compile queries into Hadoop jobs, so load level is modest, and load grows by number of job requests plus number of data inserts, multiplied by "size of data" ("number of datanodes"). If the mysql component/role is specified is a distinctive way, like hive-mysql, then the configuration reference can later be migrated to be on a separate machine. I could use a Hive service in Whirr next week, actually (a patch would be fine).
          Hide
          Andrei Savu added a comment -

          I think you can use puppet to setup the mysql server now that we have support for that in trunk.

          Show
          Andrei Savu added a comment - I think you can use puppet to setup the mysql server now that we have support for that in trunk.
          Hide
          Adam Lippai added a comment -

          I have a working Hive - Whirr integration. It needs to be configurable and tests are missing too, but I will create a patch from it today. Thank you for your patience.

          Show
          Adam Lippai added a comment - I have a working Hive - Whirr integration. It needs to be configurable and tests are missing too, but I will create a patch from it today. Thank you for your patience.
          Hide
          Andrei Savu added a comment -

          Adam any updates? I can help with adding integration tests.

          Show
          Andrei Savu added a comment - Adam any updates? I can help with adding integration tests.
          Hide
          Guido Serra aka Zeph added a comment - - edited

          Updates? I can give a hand too

          Show
          Guido Serra aka Zeph added a comment - - edited Updates? I can give a hand too
          Hide
          Russell Jurney added a comment -

          Is it possible to simply use apt-get in this recipe to install mysql-server?

          Show
          Russell Jurney added a comment - Is it possible to simply use apt-get in this recipe to install mysql-server?
          Hide
          Russell Jurney added a comment -

          Helpful hint: the way to install MySQL from console (non interactively) on Ubuntu 10.0.4 is:

          export DEBIAN_FRONTEND=noninteractive
          sudo debconf-set-selections <<< 'mysql-server-5.1 mysql-server/root_password password diurd'
          sudo debconf-set-selections <<< 'mysql-server-5.1 mysql-server/root_password_again password diurd'
          sudo apt-get -q -y -V --force-yes --reinstall install mysql-server-5.1

          Some of the flags on apt-get being optional.

          Show
          Russell Jurney added a comment - Helpful hint: the way to install MySQL from console (non interactively) on Ubuntu 10.0.4 is: export DEBIAN_FRONTEND=noninteractive sudo debconf-set-selections <<< 'mysql-server-5.1 mysql-server/root_password password diurd' sudo debconf-set-selections <<< 'mysql-server-5.1 mysql-server/root_password_again password diurd' sudo apt-get -q -y -V --force-yes --reinstall install mysql-server-5.1 Some of the flags on apt-get being optional.

            People

            • Assignee:
              Unassigned
              Reporter:
              Andrei Savu
            • Votes:
              7 Vote for this issue
              Watchers:
              8 Start watching this issue

              Dates

              • Created:
                Updated:

                Development