This project is read-only.


Definitive installation instructions are in the downloads, but they are repeated here so that you can see what you're getting into before deciding to download.

  1. Follow the instructions in Readme.txt in the CouchDb directory of the distribution to set up CouchDb. See below for these.
  2. Open the Hydra solution in the DotNet directory and compile the Messaging project.The DLLs created in its bin directory are the ones you will need to use Hydra in your projects. If you are going to use PubSubByType or Conversations, you will need to compile up those projects as well.
  3. Try the PubSubByTypeExample project:
        a) Check that project's App.config file has the right settings for your Hydra database.
        b) Set the solution's startup project to PubSubByTypeExample and run it.
  4. Try the conversation example:
        a) Check that the settings in the App.config files for ConversationExampleServer and ConversationExampleClient are correct.
        b) Set the solution's startup projects to be ConversationExampleServer and ConversationExampleClient.
        c) Run the solution.

CouchDb setup

Install CouchDb Version 1.2: get the installer from Install.

Then modify Erlang as described in Erlang\Readme.txt. This readme is not replicated here.

If this is the first instance of CouchDb, create the Hydra database as below. Otherwise go to the config step.

  1. In Futon create an admin logon with the standard username and password (admin/match in all the Hydra examples). Note that Futon may misbehave in browsers other than Firefox, so it is safest to use that.
  2. Log in as your admin user.
  3. Click Create database and enter a name for your Hydra database. Example programs in the Hydra distribution assume the database is called hydra, but this can be changed in app.config.
  4. Go into your new database and create design documents as follows:
       Click New document, then click the Source box on the resulting page.
       Replace the text with the contents of the validate.json file from the CouchDb project in the Hydra distribution.
       Click Save document.
       Repeat with the contents of the hydra.json document.


  1. In Futon, go to Configuration.
  2. Set bind_address=, max_replication_retry_count=infinity, delayed_commits=false. You may also want to set level=error in the log section, to prevent your log becoming too huge.
  3. Ensure that algorithm=utc_machine_id (this should have been done during the Erlang changes above).
  4. Ensure that machine_id is a unique value across all CouchDb instances (see the Erlang changes for a bit more on this).


If this is not the first instance of CouchDb, then set up replication.

  1. In Futon go into the _replicator database.
  2. For each machine, <machine_name> with which you want to replicate:
       a. Create a document in _replicator called PullFrom<machine-name>
       b. Give it fields create_target=true, continuous=true, source=http://admin:match@<machine_name>:5984/hydra, target=http://admin:match@
          It's probably best to make the <machine_name> in source a FQDN to prevent ambiguity.
       c. Save the document. Check that it gets modified to say replication_state=triggered.
  3. For each of the replication targets above, create an equivalent pull replication in its _replicator database. Fields will be identical except that the machine name in source will be the machine on which you're installing CouchDb.


If you are using CouchDb 1.2, you can set up regular compaction. This is a new feature and it's not yet clear which setting work best. They will also depend on the volume of Hydra traffic in your database, so you should experiment.

  1. In Futon, go to Configuration
  2. Click the "Add a new section" link at the bottom of the page.
  3. Set section=compaction, option=hydra (or whatever name you are using for your hydra database), and value=[{db_fragmentation, "70%"}, {view_fragmentation, "60%"}, {from, "23:00"}, {to, "04:00"}]
    or something similar, where the fragmentation values are percentages of old data and old view index data in the database, and from and to are times of day between which compaction should take place.

Last edited May 2, 2012 at 2:43 PM by NickNorth, version 1


No comments yet.