look into something like "sticky sessions"?
>
> I have experience in application design, but not as much with server
> configuration and clustering so I'm not sure how this is normally solved.
> Thanks in advance for your insight!
>
>
~~~
gt; I should look into something like "sticky sessions"?
There isn't really anything that rises to the level of a best practice
here. Instead, different application environments have different
desired outcomes.
Are you using clustering to support a larger number of users than a
single
experience in application design, but not as much with server
configuration and clustering so I'm not sure how this is normally solved.
Thanks in advance for your insight!
~|
Order the Adobe Coldfusion Anthology now!
http://w
that's what I meant
On Sat, Aug 13, 2011 at 3:56 AM, Dave Watts wrote:
>
> > I've used the standalone installer, but it's been a good long time. I
> > think it only accepts requests from localhost, or some such pain in the
> > arse deal.
>
> No, the localhost limitation applies to the spider.
> I've used the standalone installer, but it's been a good long time. I
> think it only accepts requests from localhost, or some such pain in the
> arse deal.
No, the localhost limitation applies to the spider. The standalone
Verity will accept requests from other IP addresses.
Dave Watts, CTO,
ah sorry yes in that case your right, verity will only search localhost, so
it is only good for 1 site unless you pay for it.
On Sat, Aug 13, 2011 at 1:36 AM, Matthew Williams
wrote:
>
> The restriction isn't for CF, I meant for Verity. It's been my
> understanding that the only way to get a fu
The restriction isn't for CF, I meant for Verity. It's been my
understanding that the only way to get a full on Verity install that
multiple servers could connect to was to pay for a fullblown license.
--
Matthew Williams
Geodesic GraFX
www.geodesicgrafx.com/blog
twitter.com/ophbalance
there is no such restriction, otherwise CF would not be much use in a
production environment if no-one could connect to it would it :-)
On Fri, Aug 12, 2011 at 10:38 PM, Matthew Williams wrote:
>
> I've used the standalone installer, but it's been a good long time. I
> think it only accepts re
I've used the standalone installer, but it's been a good long time. I
think it only accepts requests from localhost, or some such pain in the
arse deal.
--
Matthew Williams
Geodesic GraFX
www.geodesicgrafx.com/blog
twitter.com/ophbalance
~
> That's only possible if you use Verity as a standalone product (not
> free), sadly. SOLR (free!) supports this, however, but you'd need to
> convert everything over to SOLR collections. I believe there's a
> function in the admin to do that conversion.
I haven't used Verity in a while (thankf
That's only possible if you use Verity as a standalone product (not
free), sadly. SOLR (free!) supports this, however, but you'd need to
convert everything over to SOLR collections. I believe there's a
function in the admin to do that conversion.
--
Matthew Williams
Geodesic GraFX
www.geode
Thanks for the info, this is what I thought, just needed to make sure.
Is it also possible to share a verity collection, so each cluster / instance
has a single verity collection?
~|
Order the Adobe Coldfusion Anthology now!
h
Thanks for the info, this is what I thought, just needed to make sure.
Is it also possible to share a verity collection, so each cluster / instance
has a single verity collection?
~|
Order the Adobe Coldfusion Anthology now!
h
> My question is: We plan to setup 6 instances and 3 clusters (2 instances in
> each cluster). How do we assign domains to specific clusters, in
> effect 7 domains (each with differnet ip address) to a specific cluster. Is
> this done in wsconfig? Is it doable?
You'll first have to create y
Hi All
We have a cf9/windows 2003 server and we're about to install CF9 multiserver.
We use IIS6 and have 21 domains each with their own IP address (for SSL
reasons).
My question is: We plan to setup 6 instances and 3 clusters (2 instances in
each cluster). How do we assign domains to specif
Sent: 26. januar 2011 04:48
To: cf-talk
Subject: Re: Clustering ColdFusion - some "black holes"
> But, the below described procedure - it's 100% accurate for setting up a
> cluster?
> (not ticking of the "replicate sessions" this time.. :))
>
> --
fusion1", 8302 for instance two "cfusion2", 8303 for
> instance 3 "cfusion3" on physical server #2 etc.)
> Every CF instance have the remote port (JNDI port) set unique (both server
> wide and cluster wide)
>
> Clustering setup
> On one server (CFadmin / Inst
rate instance names (8300 for admin, 8301 for instance one "cfusion1",
8302 for instance two "cfusion2", 8303 for instance 3 "cfusion3" on physical
server #2 etc.)
Every CF instance have the remote port (JNDI port) set unique (both server wide
and cluster wide)
Clust
]
Sent: Tuesday, January 25, 2011 11:29 AM
To: cf-talk
Subject: RE: Clustering ColdFusion - some "black holes"
I agree also.
I should have left that one out of it..
It seems to have hidden the real question which is: "what is the 100%
correct way to create a cluster as this.&
on't have their IPs changed in
mid-visit, it would also not be needed.
From: "Sean Corfield"
Sent: Tuesday, January 25, 2011 12:48 PM
To: "cf-talk"
Subject: Re: Clustering ColdFusion - some "black holes"
I hope M
I hope Mike Brunt is still on this list and will jump in on this
because he has a lot of experience with clustering JRun and has a lot
to say about session replication and other options (basically he
agrees with Dave :)
My experience with session replication on JRun was that replication
could
I agree also.
I should have left that one out of it..
It seems to have hidden the real question which is: "what is the 100% correct
way to create a cluster as this."
Thanks,
Helge
-Original Message-
From: Jason Fisher [mailto:ja...@wanax.com]
Subject: Re: Clustering ColdFus
eb servers.
From: "Dave Watts"
Sent: Tuesday, January 25, 2011 9:11 AM
To: "cf-talk"
Subject: Re: Clustering ColdFusion - some "black holes"
> Goal:
> One rock stable high performance ColdFusion cluster with session
> Goal:
> One rock stable high performance ColdFusion cluster with session
> replication and quick failover
Is there a specific requirement for session replication? Because you might get
generally better results if you just use sticky sessions.
That buys you load balancing, but not complete fai
> Goal:
> One rock stable high performance ColdFusion cluster with session replication
> and quick
> failover
Is there a specific requirement for session replication? Because you
might get generally better results if you just use sticky sessions.
That buys you load balancing, but not complete fa
Hello,
There is some questions regarding clustering I've never found a clear /
consistent answer on.
I hope someone can give me the ultimate answers on my questions..
I've tried to generalize a bit to "broaden" the scope for this,
Software
Win2003/2008 32/64 bit -
Thanks for the tips guys, I appreciate it. This gives me a decent place to
start. I wish we could go with the Enterprise version to get some of the
extended features but it's out of our budget.
~|
Want to reach the ColdFusion
We always had good luck with hardware load balancers from Coyote Point,
managing connections to 4 or 5 clustered web servers. For apps that needed to
maintain seamless persistence, we used client vars (stored in the central DB),
which allows the application to keep a running connection to a us
been using them
for about 5 years and never had a problem. And I've always gotten great
support when I needed it...
Brook
-Original Message-
From: Dave Watts [mailto:dwa...@figleaf.com]
Sent: December-01-09 8:40 AM
To: cf-talk
Subject: Re: CF clustering with Standard version vs Enterp
> Can someone enlighten me on the best direction to go for setting up
> clustering with CF Standard versions. I realize you can do
> this with the Enterprise version but is there an operating system/web server
> or software based method to manage load balancing,
> then point to
Can someone enlighten me on the best direction to go for setting up clustering
with CF Standard versions. I realize you can do this with the Enterprise
version but is there an operating system/web server or software based method to
manage load balancing, then point to my different CF instances
I get the following error while trying to add an instance to a cluster:
No MBean is registered on server under the ObjectName *:*.
Also, trying to get to the administration server for this particular
instance results int he following HTTP 500 error:
There is no web application configured to se
Matthew,
That is excellent.
By any chance, is there a nice technical article available online that we can
look into regarding NFS clustering?
On the side, customer is saying that proper clustering should take care of code
sync. So, if I push a file on the CF/JRun4 it should automatically
, 2009 at 1:25 PM, Matthew Williams
wrote:
>
> That's going to highly depend on your server architecture. We have
> Windows 2003 R2 Enterprise. The NFS clustering in that release has been
> fantastic, and has only failed one time in about a year. It's
> multi-point sync
That's going to highly depend on your server architecture. We have
Windows 2003 R2 Enterprise. The NFS clustering in that release has been
fantastic, and has only failed one time in about a year. It's
multi-point sync (a -> b, b -> a) and can do partial file changes as
If you're on Windows 2003 R2 you can use NFS replication for this.
-Helge
-Original Message-
From: Joseph Bugeja [mailto:jbug...@rs2group.com]
Sent: 18. februar 2009 19:17
To: cf-talk
Subject: Clustering with Synchronization/Mirroring
We have a setup consisting of 2 CF8 se
We have a setup consisting of 2 CF8 servers running in parallel (the 2
instances are clustered together using CF8 Ent/JRun4) and we have one app.
deployed on each of the two servers. Each request is served by either one of
the 2 servers.
Although clustering is working great, we would like
As it turns out this was a coincidence. It still takes half an hour for
members in different datacentres to join the cluster after a restart;
members in the same datacentre join as soon as the instance is started.
mxAjax / CFAjax docs and other useful articles:
http://www.bifrost.com.au/blog/
20
ocs and other useful articles:
> >> > http://www.bifrost.com.au/blog/
> >> >
> >> >
> >> > 2008/10/2 AJ Mercer <[EMAIL PROTECTED]>
> >> >
> >> > > for point 2there maybe some timeout setting on the network - if it
> >
; >
>> > > for point 2there maybe some timeout setting on the network - if it
>> > > cant
>> > see
>> > > it for a period of time it may think the cluster is dead.
>> > >
>> > > Is the newtowrk under load when it drops out?
>>
ng on the network - if it cant
> > see
> > > it for a period of time it may think the cluster is dead.
> > >
> > > Is the newtowrk under load when it drops out?
> > >
> > >
> > > On Thu, Oct 2, 2008 at 2:28 PM, James Holmes <[EMAIL PROTECTED
be some timeout setting on the network - if it cant
> see
> > it for a period of time it may think the cluster is dead.
> >
> > Is the newtowrk under load when it drops out?
> >
> >
> > On Thu, Oct 2, 2008 at 2:28 PM, James Holmes <[EMAIL PROTECTED]
TED]>
> for point 2there maybe some timeout setting on the network - if it cant see
> it for a period of time it may think the cluster is dead.
>
> Is the newtowrk under load when it drops out?
>
>
> On Thu, Oct 2, 2008 at 2:28 PM, James Holmes <[EMAIL PROTECTED]>
> wr
for point 2there maybe some timeout setting on the network - if it cant see
it for a period of time it may think the cluster is dead.
Is the newtowrk under load when it drops out?
On Thu, Oct 2, 2008 at 2:28 PM, James Holmes <[EMAIL PROTECTED]> wrote:
> We have two problems with clust
We have two problems with clustering our CF 8.01 Multiserver (i.e. JRun)
installs.
1) Session replication is broken. When we turn off an instance in a cluster
that has session replication and stickiness turned on, the next server that
gets the request throws a null pointer error when trying to
Here is a blog entry that details the configuration for Apache (disclaimer:
I wrote it):
http://speeves.erikin.com/2007/01/coldfusion-clustering-faq.html
To manually run the connector:
*Tip: *(Server configuration only) To use the command line, open the batch
files located in *cf_root*\bin
Thanks for the suggestion. Not a firewall issue. All the instances are on the
same machine, and the port for the instances is open between the DMZ and our
internal lan.
I'm almost positive it has something to do with the Web Configuration, but
other than the Web Server Configuration Tool, I d
What happens when you shutdown the master node? Could it be a firewall
setup which is blocking communication between nodes?
On Fri, Sep 12, 2008 at 8:11 AM, Scott Taylor <[EMAIL PROTECTED]>wrote:
> I have a CF cluster I just set up with three instances - Master, and two
> additional. When I sta
I have a CF cluster I just set up with three instances - Master, and two
additional. When I start the additional instances, the logs (*-out.log) show
that the instances see the cluster and join fine.
However, when I run the server monitor for each instance, I only see active
sessions going to
> Does anyone do CF clustering with remote instances? How does
> that work?
>
> I mean... your web root exists on Server A, but you have CF
> instances on SErver A and Server B that do the processing.
>
> If you do things like createobject or cfinclude, or cffile...
>
the remote server would
probably have to have the same directory structure and any cffile write
actions would have to go to some kind of shared folder or be replicated
between servers.
Rick
On Fri, Jun 13, 2008 at 8:42 AM, Rick Root <[EMAIL PROTECTED]>
wrote:
> Does anyone do CF clu
Does anyone do CF clustering with remote instances? How does that work?
I mean... your web root exists on Server A, but you have CF instances on
SErver A and Server B that do the processing.
If you do things like createobject or cfinclude, or cffile... would server B
have to access the file
we have unique ports per server (that is port 8301 & 8302 is used on two
server), but we are having issues adding a third server to the cluster - I
too would be interested in hearing a authoritative view on this question.
On Sun, Apr 20, 2008 at 1:30 AM, WebSite CFtalk <[EMAIL PROTECTED]> wrote:
lk
Subject: Re: ColdFusion 8.0 clustering - quick question.
I think one requirement for session replication is to turn on J2EE
sessions - do you have that on?
On Sat, April 19, 2008 13:30, WebSite CFtalk wrote:
> Hello,
>
> I'm struggling with a cluster setup. (IIS6/CF8, 2 servers
You also need to name your instances uniquely. Even if you just copied
the ear/war file, the instance names in the Enterprise Admin (JRun
admin) cannot match. Just another thing to check.
Matthew Williams
Geodesic GraFX
www.geodesicgrafx.com/blog
~
I think one requirement for session replication is to turn on J2EE
sessions - do you have that on?
On Sat, April 19, 2008 13:30, WebSite CFtalk wrote:
> Hello,
>
> I'm struggling with a cluster setup. (IIS6/CF8, 2 servers with 2
> instances each)
>
> The problem is session replication - tons of e
Hello,
I'm struggling with a cluster setup. (IIS6/CF8, 2 servers with 2
instances each)
The problem is session replication - tons of errors regarding "session
replication failed" in the logs:
" 19/04 19:01:40 error Setup of session replication failed.
[2]java.io.StreamCorruptedException: unexpec
properly?
> In our situation, clustering across physical servers won't really help us
> because our networking staff has our load balancer permanently directing
> blocks of users to one server or the other.
That doesn't matter to CF, servers on the webserver on server 1
tance, how are you notified if one of the
> instances in your cluster is failing?
Just figured I'd post some results, as I have been playing with CF8 clustering
in our development environment.
I must say that I am extremely pleased so far with CF8's ability to failover
from a dead in
>The Client scope works fine for that, actually, as long as you're using a
>database. It's relatively slow, but it's reliable.
Yeah that would be my concern, if I am scaling my app by clustering the last
thing I would think I would want to do is slow it down with constant
It depends on the number of queued threads and the default stack size.
Some platforms have a 512k default stack, for example. Queue 100
threads and that's 50MB, because the thread is assigned the stack when
it's queued, not when it's executed.
As a comparison point, we often have a few hundred con
se you might be able to use client
> scope if you dared, never tried).
The Client scope works fine for that, actually, as long as you're using a
database. It's relatively slow, but it's reliable.
Again, though, clustering multiple instances on a single physical server
isn't s
> Sort of. Sessions certainly take up the same room in both if
> you're using session replication and you'd expect that over
> time the same set of classes are going to be created and take
> up perm space. But since the threads are spread between each
> instance there's (approximately) half the
> You don't necessarily have the same stuff in both instances,
> as that only replicates sessions. You could have different data
> cached in your application scope based on which clients hit that
> server.
You could, but that would be both unlikely and problematic, since your
clients would be
> -Original Message-
> From: Dave Watts [mailto:[EMAIL PROTECTED]
> Sent: Monday, March 17, 2008 8:48 PM
> To: CF-Talk
> Subject: RE: CF 8 Instance Clustering: for performance? or failover?
>
> > It definitely improves things for us. Not only does it give
>
Sort of. Sessions certainly take up the same room in both if you're
using session replication and you'd expect that over time the same set
of classes are going to be created and take up perm space. But since
the threads are spread between each instance there's (approximately)
half the stack space b
That all depends on how you configure the cluster. As long as replication is
not on then no you would not be storing the same data, so you could get more
out of your memory, at that point failover would only keep folks online the
session would be lost (though I suppose you might be able to use clie
> It definitely improves things for us. Not only does it give
> you an extra instance in case something goes wrong with the
> original one, it allows you to utilize 2GB of ram instead of 1GB.
That doesn't improve performance, just reliability. And if you cluster two
instances, you're storing th
age-
> From: Jochem van Dieten [mailto:[EMAIL PROTECTED]
> Sent: Monday, March 17, 2008 4:05 PM
> To: CF-Talk
> Subject: Re: CF 8 Instance Clustering: for performance? or failover?
>
> Steven Wood wrote:
> > With multi-server CF8 you can duplicate an instance and then link them
Steven Wood wrote:
> With multi-server CF8 you can duplicate an instance and then link them
> together as a cluster on the same machine. Does this improve performance in
> any way?
It shouldn't. (Unless the limiting factor is Java heap space and you get
better performance from being able to pa
With multi-server CF8 you can duplicate an instance and then link them together
as a cluster on the same machine. Does this improve performance in any way?
Or is it purely for failover should 1 of the instances halt in any way?
And with the failover circumstance, how are you notified if one of
James Buckingham wrote:
> The log file was from the failing instance. It was residing in the
> C:\JRuns\Logs\ folder.
Is there anything from the "Clusterable service" in your *-event.log on
the failing or other instances? At the very least you should see some of
it when the instance starts or s
Right here's hoping that this doesn't start another thread as this is getting
embrassing!!!
That second set are logs are obviously related to the drop of the instance and
JRun starting to hand the sessions across to the other server. Not very helpful
but still more information I can work on.
Y
diagram stuck in my head that
was posted yesterday on the list. :)
~Brad
-Original Message-
From: James Buckingham [mailto:[EMAIL PROTECTED]
Sent: Thursday, February 28, 2008 9:30 AM
To: CF-Talk
Subject: Re: JRun Clustering problem - Instance keeps "shutting
down"
The log fil
And why do I keep getting my replies in separate entries :-(
Sorry people!!
> The log file was from the failing instance. It was residing in the
> C:\JRuns\Logs\ folder.
>
> IIS is sitting on both boxes so I'm not sure if that classes as a
> "dedicated IIS box". We have round robin on the DN
The log file was from the failing instance. It was residing in the
C:\JRuns\Logs\ folder.
IIS is sitting on both boxes so I'm not sure if that classes as a "dedicated
IIS box". We have round robin on the DNS as well so requests are being passed
to both boxes.
I actually didn't know about those
From: Brad Wood [mailto:[EMAIL PROTECTED]
Sent: Thursday, February 28, 2008 8:21 AM
To: CF-Talk
Subject: RE: JRun Clustering problem - Instance keeps "shutting
down"
Is it possible there is a connectivity issue which is causing the JRun
connectors (which are loaded on the IIS box and c
James Buckingham [mailto:[EMAIL PROTECTED]
Sent: Thursday, February 28, 2008 7:17 AM
To: CF-Talk
Subject: Re: JRun Clustering problem - Instance keeps "shutting
down"
Thanks AJ. Is that the "give it a kick" approach :-)
We actually had a phase before all this started whe
That is probably because the JRun connector loaded in IIS is what
divvies up the requests.
~Brad
-Original Message-
From: AJ Mercer [mailto:[EMAIL PROTECTED]
Sent: Thursday, February 28, 2008 6:38 AM
To: CF-Talk
Subject: Re: JRun Clustering problem - Instance keeps "shutting dow
Thanks AJ. Is that the "give it a kick" approach :-)
We actually had a phase before all this started where IIS stopped processing
requests and then "very kindly" jammed up both server instances, stopping the
site completely. It wasn't until we restarted it or the server that it would
release t
Thanks AJ. Is that the "give it a kick" approach :-)
We actually had a phase before all this started where IIS stopped processing
requests and then "very kindly" jammed up both server instances, stopping the
site completely. It wasn't until we restarted it or the server that it would
release th
backup software on the other "working" server. So I wouldn't expect this to
> effect the one that is now faulty.
>
> We started to add Hotfixes to the server on the 5th of Feb but this hasn't
> resolved the problem. These included JRun Updater 7 and CF HotFix 3
pdater 7 and CF HotFix 3 which had
stability fixes for clustering. This was applied to both servers.
Checking logs in IIS I'm not seeing anything abnormal.
We've compared the setup of the failing instance to the other server instance
and apart from a few unique configuration setting
pdater 7 and CF HotFix 3 which had
stability fixes for clustering. This was applied to both servers.
Checking logs in IIS I'm not seeing anything abnormal.
We've compared the setup of the failing instance to the other server instance
and apart from a few unique configuration setting
ber 12, 2007 9:53 PM
To: CF-Talk
Subject: Re: clustering issue
In addition to that, you might want to check out Sarge's blog and do
some jrun.xml tuning. The problem with relying on CF to time out
requests is that it doesn't work for third party calls. So, what's a
th
In addition to that, you might want to check out Sarge's blog and do
some jrun.xml tuning. The problem with relying on CF to time out
requests is that it doesn't work for third party calls. So, what's a
third party call, you might ask? Would you think that DB connections
would count as third
i think what happened is that requests were piling up and not timing
out and so I need to get Fusion Reactor back on my instances for
crash protection.. maybe that'd help :)
Fusion Reactor worked wonders in Cf7 and I just haven't installed it
since we migrated to CF8...
Rick
I've actually been impressed by the JRun clustering of CF7. Most oft
times what fails for me is IIS stops responding, but my CF instances are
fine. Our load balancer only checks to see if the HTTP port can be
opened, not that it gets back a correct response. The switch people
tell me t
clusters is
> not responding and direct the requests to the other?
How long did this situation exist? If it was going on for a longer
period the clustering should know one of the instances wasn't responding
and should stop directing requests to it (may take some time depending
on where
Brad is dead on here.
JRun clustering is not the best solution for most things. There are a few
edge cases where it can be helpful, especially if you're willing to spend
money, but generally, I would recommend a hardware load balancer like an F5.
--
nathan strutz
http://www.dopefly.com/
O
If CF8 clustering is like CF7, I believe you are giving it too much
credit.
The JRUN connectors set up in your web server will not give request to
an instance which has been stopped, but it pays absolutely no regard to
the "load" of that instance.
I was disappointed when I foun
topped the errant instance,
requests went through to the other instance just fine.
Why is this? Shouldn't the listener know that one of the clusters is
not responding and direct the requests to the other?
Or am I giving the coldfusion clustering capabilities way too much credit.
My inst
are answering on a context root of /. I
> am using connector clustering to load balance requests coming in from
> my single instance of Apache. The first time I set this up I created
> an ear file using the packaging and deployment utility, then created
> my 1st instance and deployed it f
On Tuesday 04 Dec 2007, Andy Cortright wrote:
> Is it ok to have Apache Docroot set to the application server webroot of
> the 1st instance? What are my options and what are generally accepted best
Seems sensible.
It's what we did when we bothered with multi-server before giving it up as too
much
/. I am using
connector clustering to load balance requests coming in from my single instance
of Apache. The first time I set this up I created an ear file using the
packaging and deployment utility, then created my 1st instance and deployed it
from the ear file (repeated for each additional
> Can a coldfusion request know the name of the instance it's
> running in?
Yes:
http://www.sumoc.com/blog/index.cfm?mode=entry&entry=7B541DBA-5004-2066-B7BC
F2D895B3EC4A
Dave Watts, CTO, Fig Leaf Software
http://www.figleaf.com/
Fig Leaf Software provides the highest caliber vendor-authorize
Dave Watts wrote:
>
> Set a flag or flags somewhere that all the servers can see, and have them
> check the flag or flags periodically. Alternatively, have each server send
> an HTTP request to the others, to have them update their cache. This kind of
> thing is well-suited for the event gateway,
> In my web app, I load a lot of stuff that I access often into
> the application scope - numerous queries and such.
>
> When I update the data behind those queries, I just pass
> something like
> init=1 to the main url, and that triggers reloading of those
> queries into the application scope.
In my web app, I load a lot of stuff that I access often into the
application scope - numerous queries and such.
When I update the data behind those queries, I just pass something like
init=1 to the main url, and that triggers reloading of those queries
into the application scope.
However, now
Dave Watts wrote:
>
> If both cluster instances are on the same physical machine, and both
> instances will be connected with a single virtual web server, you don't need
> to copy your CFM files. You don't even need to go through creating and
> deploying an EAR file at all; it's just a simple way
1 - 100 of 288 matches
Mail list logo
