On Jun 2, 2008, at 6:20 PM, Aristedes Maniatis wrote:
On 03/06/2008, at 10:17 AM, Chuck Hill wrote:
I am not sure that this is going to work with wotaskd and friends.
What does hostname show on this machine? serverb.internal or
something else? If something else, that is probably the name you
want to be using. Unless it says XXXX.local, which you really
don't want to use. :-)
Thanks for your help Chuck.
To get this straight, are you saying that:
unix hostname == WOHost == Javamonitor host name
My experience has been that you run into the fewest problems with a
configuration like that.
That is, we should be setting these three things to match each
other. If we have two application servers, then WOHost should be set
for each instance to point to the hostname of the machine it is
running on, regardless of where Apache is running.
And wotaskd should probably have the same WOHost parameter - just in
case. Ideallly, just making the host name in JavaMonitor match the
hostname of the name should be sufficient.
The docs for WOHost say "The name of the host on which the
application is running or, in a multihomed environment, the address
of the adaptor the application should use. See the host method of
WOApplication for details." I don't quite get the 'address of the
adaptor' part. It seems to indicate something about the URL or the
machine Apache is running on, but I don't understand how that makes
any sense.
It does not make any sense to me either. I also always set the
WOCGIAdaptorURL property and that has the externally visible host name
(e.g. where Apache is running).
Since we want to keep WebObjects hidden away from the world on
private IP addresses (which are networked between servers in the
cluster on a separate gigabit switch and separate NICs), then we've
mapped serverA.internal and serverB.internal to those private IP
addresses. It sounds like this isn't possible under WO without also
having the machine's hostname pointing to the same private host/IP.
This is all sort of voodoo to me. I strongly suspect the related code
is not the finest in WebObjects. You might be able to make it work by
using the IP address in JavaMonitor.
Is the above a fair summary of this position? Do the new Project
Wonder versions of wotaskd/javamonitor suffer from these same
limitations?
I think that is correct. That is how it has appeared to work (or not
work, depending on your point of view). I know that Anjo fixed and
extended these a lot in the Wonder versions. I don't recall the
specifics and have not started to use them yet.
Chuck
--
Practical WebObjects - for developers who want to increase their
overall knowledge of WebObjects or who are trying to solve specific
problems.
http://www.global-village.net/products/practical_webobjects
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-deploy mailing list (Webobjects-deploy@lists.apple.com)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-deploy/archive%40mail-archive.com
This email sent to [EMAIL PROTECTED]
