Yeah, the more I dug into this I realized this is how clustercats was
monitoring the load state of each machine in the cluster... I verified that
/btauxdir/cfprobe.cfm was public viewable over http://.
I went ahead and took a look at the log files for each web server and I can
see where ever 60 seconds or so a hit is made to cfprobe.cfm with a browser
agent of Allaire+ClusterCATS+probe. I can even see my own hits to the page
as well hitting inbetween clusterCATS polling the state. Unfortunatly there
is no apparent error happening. =(
Anyone else have any ideas?
Trey
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]
Sent: Thursday, September 13, 2001 7:18 AM
To: CF-Talk
Subject: RE: Trouble with ClusterCats
I had a problem with similar symptoms; turned out that the probe process
couldn't access cfprobe.cfm. The method ClusterCats uses to decide where to
look for cfprobe.cfm is not clear to me, and if it can't get it for any
reason it decides that the server is busy. Look in your IIS (or whatever)
logs for requests to cfprobe.cfm and see if they are being denied for some
reason.
> -----Original Message-----
> From: Trey Rouse [mailto:[EMAIL PROTECTED]]
> Sent: Wednesday, September 12, 2001 7:01 PM
> To: CF-Talk
> Subject: Trouble with ClusterCats
>
>
> I'm having some troubles configuring clustering on CF5.
>
> Specificaly, the load monitor on all servers runs at less
> than 3% for 20-30
> seconds then shoots up to 100% and makes the machine in the
> cluster offline.
>
> To simplify testing this, the problem duplicates by adding
> only a single
> machine to the cluster and bringing up load monitor.
>
> Here is my process:
>
> 1. Create the Cluster
> 2. Add A SINGLE machine to the cluster
> 3. Configure a basic probe to the btauxdir\cfprobe.cfm file
> checking for
> "Hello"
> 4. Manually verify cfprobe.cfm is working
> 5. Bring up the load monitor on the single server.
>
> Within 30 seconds server load is reported at 100%. Using OS tools and
> checking the service, there is no noticable load.
>
> I have found restarting the CF monitoring service will return state to
> normal for another 20-30 seconds before jumping up to full
> utilization.
>
> I'm using a very simple configuration, 2 identical servers
> set up on DNS
> round robin. No hardware ballancing. My goal is primarily failover
> protection for when one machine becomes unavailable.
>
> I did find the hotfix about CPU utilization. I didn't think
> it would apply
> to this as OS CPU utilization is unaffected by this, but the results
> duplicate regardless of applying the hotfix or not.
>
> On an unrelated note, I've descovered on every install of
> every machine I've
> ever made of CF5 w/ custerCATS the reset feature under ClusterCATS
> Administrator -> Advanced -> Reset fails with an unable to
> remove Btauxdir
> error message.
>
> I've looked around and found no related issues reported in
> support forums or
> addressed in a knowledge base article (go figure) so I
> thought I would see
> if anyone has any ideas here...
>
> Trey Rouse
> [EMAIL PROTECTED]
>
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Get the mailserver that powers this list at http://www.coolfusion.com
FAQ: http://www.thenetprofits.co.uk/coldfusion/faq
Archives: http://www.mail-archive.com/[email protected]/
Unsubscribe: http://www.houseoffusion.com/index.cfm?sidebar=lists
