Hi Alan,

Thanks for pointing me in the right direction. 

I think I have been getting caught up in the terminology and the
definitions of active/active. 

You are right I have sorted out the active/passive part of the ha
offering.

I have been reviewing the documentation with regards to clones
(http://linux-ha.org/ClusterInformationBase_2fClones). This is posing me
a big problem.

When migrating my previous test apps to use clones, they seem to
monitor, stop but never startup. I am getting some errors with my
heartbeat script in /etc/ha.d/resource.d. This script worked fine before
implmenting clones, which raiswes the question, does clones support
primitives with class of heartbeat?

Also, I have tried the example you had on the documentation pages and
that appears to not be working either. Crm_mon shows that none of the
primitives are active. I have provided  my simple setup below. Can you
see anything blatently wrong with them, or do you have a complete
working example you could send me?

Many thanks,

Bob


Here are my setup files:

==================== Ha.cf ==============================
use_logd on
udpport 694
keepalive 1
deadtime 10
bcast eth0
node sid.nexagent.com nancy.nexagent.com
crm yes
==================== Cib.xml ============================
 <cib>
   <configuration>
     <crm_config>
     </crm_config>
     <nodes/>
     <resources>
<clone id="MyCorpWWW">
  <instance_attributes>
    <attributes>
      <nvpair name="clone_max" value="3"/>
      <nvpair name="clone_node_max" value="1"/>
    </attributes>
  </instance_attributes>
  <group id="MyCorpWWW_group">
    <primitive id="WebServerIP" class="ocf" type="IPaddr"
provider="heartbeat">
      <instance_attributes>
        <attributes>
          <nvpair name="ip" value="192.168.13.66"/>
        </attributes>
      </instance_attributes>
    </primitive>
    <primitive id="WebServerApache" class="ocf" type="Apache"
provider="heartbeat">
      <instance_attributes>
        <attributes>
          <nvpair name="apache_config"
value="/etc/httpd/conf/httpd.conf"/>
        </attributes>
      </instance_attributes>
    </primitive>
  </group>
</clone>
     </resources>
     <constraints>
     </constraints>
   </configuration>
   <status/>
 </cib>
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of
[EMAIL PROTECTED]
Sent: 05 December 2005 19:00
To: [email protected]
Subject: Linux-HA-Dev Digest, Vol 25, Issue 6

Send Linux-HA-Dev mailing list submissions to
        [email protected]

To subscribe or unsubscribe via the World Wide Web, visit
        http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
or, via email, send a message with subject or body 'help' to
        [EMAIL PROTECTED]

You can reach the person managing the list at
        [EMAIL PROTECTED]

When replying, please edit your Subject line so it is more specific than
"Re: Contents of Linux-HA-Dev digest..."


Today's Topics:

   1. Re: Re: Active Active Cluster (Alan Robertson)
   2. Re: Re: [Linux-ha-cvs] Linux-HA CVS: lib by sunjd from
      (Alan Robertson)


----------------------------------------------------------------------

Message: 1
Date: Mon, 05 Dec 2005 09:27:30 -0700
From: Alan Robertson <[EMAIL PROTECTED]>
Subject: Re: [Linux-ha-dev] Re: Active Active Cluster
To: High-Availability Linux Development List
        <[email protected]>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Robert Szabo wrote:
> Hi Alan,
> 
> I should probably state my overall objective for Linux-ha in my 
> environment.

Thanks.  This helps a lot.

> Essentially, I plan to have multiple nodes running linux-ha. There 
> will be probably seven apps that I wish to manage on these two boxes.
> 
> 
> 4 apps which run in an active/passive fashion, the other 3 will run 
> active/active but with a twist.  These three apps will always run on 
> both nodes but will never need to tranfer from one node to another, as

> the load balancing is handled by the calling clients. What I am 
> attempting to do is use Linux-ha to provide the framework for 
> start/stop/monitoring only,  as well as restarting the app if it 
> should happen to fail.

Now I know how you are using the term active/active.

I take it you know how to handle the active/passive case already.

There are two ways you could do this.  The most "natural" way is to have
  a clone resource for your application.  I worry a little about how
your clients reach your active/active applications.  How do they address
the machines that run these services?  Do they talk to the "real" IP
addresses, or to floating (virtual) addresses that the CRM manages?

I may know what's going on.  Your two resources are identical to each
other, but not clones.  I think this is confusing some recent (and I
think misguided) changes to the CRM - which try and see what's running
before it starts.  Since it can't distinguish which copy is running -
since there is no difference, then I suspect it gets confused.

If you start these services as clone resources with clone_max=2 and
clone_node_max=1, then it will run no more than 2 copies, with no more
than 1 running on any given node.


> This is something that Veritas VCS provides for me today, however, the

> cost of VCS is extravagant to say the least. VCS does this by allowing

> you to specify which nodes are relevent to the resource group at the 
> service group level.

After you get this running, I'd love to hear a comparison between VCS
and Linux-HA.

> Version 2 of Linux-HA is a major upgrade from version 1 and so far it 
> looks very promising.

Thanks!

-- 
     Alan Robertson <[EMAIL PROTECTED]>

"Openness is the foundation and preservative of friendship...  Let me
claim from you at all times your undisguised opinions." - William
Wilberforce


------------------------------

Message: 2
Date: Mon, 05 Dec 2005 10:04:03 -0700
From: Alan Robertson <[EMAIL PROTECTED]>
Subject: Re: [Linux-ha-dev] Re: [Linux-ha-cvs] Linux-HA CVS: lib by
        sunjd from
To: High-Availability Linux Development List
        <[email protected]>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Sun Jiang Dong wrote:
> Somebody think the logs appear too frequently.
> http://www.osdl.org/developer_bugzilla/show_bug.cgi?id=985
> 
> I think the logs are not meaningful also. Anyway, it's better if there

> are multiple levels for log output in lib. e.g. cl_log2?

For some reason, I think they may have been coming out even when debug
isn't enabled.

I wonder if the logging code is correct?

And, if you want multiple levels of debug (which is a good idea, I
think), then the right thing to do is to have it be capable of being
enabled at run-time, NOT by changing the code.

-- 
     Alan Robertson <[EMAIL PROTECTED]>

"Openness is the foundation and preservative of friendship...  Let me
claim from you at all times your undisguised opinions." - William
Wilberforce


------------------------------

_______________________________________________
Linux-HA-Dev mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev


End of Linux-HA-Dev Digest, Vol 25, Issue 6
*******************************************
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/

Reply via email to