The /tmp/hostx_1234 files were indeed causing this; deleting them fixed the 
problem. Many thanks!

Ari Kuukka

  ----- Original Message ----- 
  To: General Discussion for the Resin application server 
  Sent: Tuesday, December 11, 2007 10:07 AM
  Subject: Re: [Resin-interest] Resin cluster/mod_caucho only using four 

  The ResinConfigServer is only for the clustering configuration, it doesn't 
configure the cluster itself.  In other words, mod_caucho needs to contact a 
backend server to find out what the resin.conf looks like.  The backend server 
will send the <srun> data back to mod_caucho.  

  So you really only need a single ResinConfigServer (two if you want a backup).

  I'm not sure why only 4 servers show up.

  If you turn on logging for the backend ResinConfigServer to "finer" (and 
choose a single ResinConfigServer so you know which machine is being used).  
That debug data will show the srun data going back to mod_caucho, which should 
help track down the issue.

  It's possible the mod_caucho cache is getting stuck, e.g. if you just added 
two new servers.  You might need to remove the cache file at /tmp/hostx_1234.

  -- Scott

  Ari Kuukka <[EMAIL PROTECTED]> wrote:

    I am trying to set up a Resin (3.0.24) cluster, with Apache/mod_caucho load 
balancing requests to several Resin nodes. Cluster of four Resin nodes is 
working fine, but I can't get mod_caucho to use more than four Resin cluster 

    If using for example six nodes, with Apache config:

            ResinConfigServer instance1 1234
            ResinConfigServer instance2 1234
            ResinConfigServer instance3 1234
            ResinConfigServer instance4 1234
            ResinConfigServer instance5 1234
            ResinConfigServer instance6 1234

    And Resin config:

        <http server-id="inst1" host="instance1" port="2345"/>
        <http server-id="inst2" host="instance2" port="2345"/>
        <http server-id="inst3" host="instance3" port="2345"/>
        <http server-id="inst4" host="instance4" port="2345"/>
        <http server-id="inst5" host="instance5" port="2345"/>
        <http server-id="inst6" host="instance6" port="2345"/>

          <srun server-id="inst1" host="instance1" port="1234" index="1" 
          <srun server-id="inst2" host="instance2" port="1234" index="2" 
          <srun server-id="ínst3" host="instance3" port="1234" index="3" 
          <srun server-id="inst4" host="instance4" port="1234" index="4" 
          <srun server-id="inst5" host="instance5" port="1234" index="5" 
          <srun server-id="inst6" host="instance6" port="1234" index="6" 

    The http://server/caucho-status page displays correctly all the six 
instances in the "Configuration Cluster" section, but in the "Default Virtual 
Host" only the first four nodes are listed. When checking the Resins' statuses, 
only those four first instances are getting requests forwarded to them.

    Am I missing some configuration parameter here, or doing some other obvious 
mistake? I assume more than four nodes should be possible? Many thanks for help.

    Ari Kuukka

    resin-interest mailing list


  resin-interest mailing list
resin-interest mailing list

Reply via email to