Jrun Connector Problem

[sanaullah]

Hi CFMX 7 -- MULTISERVER INSTALLATION The two instance has been working fine
for two days now i lost one instance from IIS6 connector file. I have two
instances cfusion2 + cfusion3 on one machine, when i test cluster it always
point to cfusion2... I have loose IPaddress:JCP port in jrunserver.store file
of IIS6 connector. I added IPaddress:JCP port reboot the machine, after
rebooting jrunserver.store file have only one IPaddress:JCP port for cfusion2
instance. Any one have idea how to solve this probelm.

[ke4pym]

Why are you trying to edit this manually instead of using the Web Server Configuration Tool?

[sanaullah]

Hi

web server configuration tool does not add both IPaddress:JCP port to jrunserver.store file.

I am bit confused why cfmx7 with jrun does this......

[ke4pym]

Are you running two seperate instances? Or have you clustered them in JRun? If you've not clustered them, you need to create a new cluster. Then rerun the Web Server Config Tool.

[sanaullah]

Hi I have both instance on same machine and instance are created by cfmx admin
screen. SERVER-INF/jrun.xml file have deactivated=false entry for both
instances. cluster is set-up through cfmx admin screen, reboot the machine and
rerun the web config tool but still only one instance is active in a cluster.
very strange in cfmx7 getting very difficult to set up cluster on one machine.

[ke4pym]

Hrm. I've not used the instance automagic config thing in 7. I setup my
instances and cluster the old school manual way.

When you start up the Web Server Configuration Tool, do you see cfusion2 and
cfusion3? Or do you see a cluster name? If you're seeing cfusion2 and 3 then
a cluster has not been created (they'll go away from the list if one has).

Can you log into the JRun administrator to verify whether or not a cluster has
been created?

I know multiple instance setup can be frustrating. But it is well worth the
frustration. Especially if you can learn to manupliate the instances yourself.

[sanaullah]

thanks Ke4ypm, I did make sure that i choose cluster name in web config tool.
I selected cluster-name and checked ALL IIS site for cfmx. after this i
rebboot the machine. still i have only one instance is active in a cluster.
is there any specific problem with cfmx7 with jrun for clustering and load
balancing...?

[ke4pym]

Go to IIS and lookup the mapping for .cfm (it should be something like
\jrun4\lib\wsconfig\3\). Then, goto that folder and copy the contents of your
jrunserver.store file. Change the IP addresses to x.x.x.x if you want to
protect your IP addys.

When you look in the JRun administrator, are both instances of ColdFusion
started?
In your services control panel, do you have an entry for each CFusion instance?

[sanaullah]

Hi I reinstall cfmx7 with jrun, create two instance cfusion2+cfusion3, create
cluster 'web-cluster'. Restart server. configure IIS6 with web config tool
[select web-cluster]. restart server, after all this again only cusion2 is
active in a cluster. I have make sure that every setting is correct for IIS6 ,
made changes in jrun.xml for deactivated=false, jrunserver.store file have
only cfusion2 IPaddress and not for cfusion3 instance. I tried every thing
which i discuss here, but no luck to get two instance cluster working. If any
one faced this problem please help me out, Thanks

[einfo]

I'm seeing a similar problem which we hoped would be fixed when we did our
upgrade to CFMX 7 last night. I uninstalled CF 6.1 Ent and then installed a
fresh copy of CF 7 Ent on ServerA hoping it would resolve the problem.
Running on the same subnet is another CF 7 Ent server (ServerB) that has a
Coldfusion cluster on it called cfCluster. When I did the upgrade on ServerA I
setup a cluster with the same name (cfCluster) not thinking about the other
server being out there. When I did that the 2 CF instances on ServerA found
the cluster of the same name on ServerB and set themselves up as replication
peers. The 2 instances on ServerA also found each other and the cluster worked
just fine. I don't really want the two machines to be working together at this
point so I figured I would create a cluster with a different name on ServerB so
they wouldn't be conflicting. Once I did that, the cluster on Server B still
works just fine but now the instances on ServerA won't cluster together and are
showing the same behavior seen below. Obviously the clustering can work but
for some reason the instances on ServerA can't see each other without the help
of an external server. Very strange. Ring any bells of where I could start
looking? Stephen

[ke4pym]

einfo-->How are you creating your instances/cluster? Are you using the
automagic tool? Or creating them by hand?

How do you know the two instances aren't running/working together? And to
that question goes to you, sanaullah, as well.

Some copy'n and pasting of your server.log files (from both instances) would
be helpful.

[einfo]

On 6.1 I was doing the clustering by hand and on 7 I tried the automagic tool
hoping it might fix something I missed. Both come back with the same results.
I'm know their not working together because I output the name of the JRun
server that served the page and the site only uses the one JRun server instead
of alternating. The log files when I had the cluster going between two
physical servers looked like this: Starting Macromedia JRun 4 (Build 84683),
cf3 server Warning: file url in codebase component may cause problems
(file:/D:/JRun4/lib/jrun.jar!/reggie.jar) 04/05 23:22:03 info ClusterManager:
JINI transient lookup service starting (cluster broadcast may require a
moment)... 04/05 23:22:04 info ClusterManager: JINI transient lookup service
started 04/05 23:22:05 info Watching directory,
D:\JRun4\servers\cf3\SERVER-INF\cluster. 04/05 23:22:05 info JRun Naming
Service listening on *:2908 04/05 23:22:05 info Clusterable service
JNDIContextManager discovered ClusterManager on 128.187.21.200:4160, cluster
domain: cfCluster (Lookup Service's JINI-ID:
050b2a35-def8-4b87-8cb2-ada74c7c3ba9) 04/05 23:22:05 info Clusterable service
JNDIContextManager discovered ClusterManager on 128.187.21.200:4870, cluster
domain: cfCluster (Lookup Service's JINI-ID:
f8440dd1-b0cd-4f23-ac4f-9a2a6acee947) 04/05 23:22:05 warning No sessionSecret
has been specified in jrun.xml. Installing a self generated sessionSecret.
04/05 23:22:05 info JNDIContextManager successfully joined a JINI lookup
service (assigned JINI-ID f8e83db9-5888-4a97-8ed1-a13636bbd7f1) 04/05 23:22:05
info Clusterable service JRunRMIBroker discovered ClusterManager on
128.187.21.200:4870, cluster domain: cfCluster (Lookup Service's JINI-ID:
f8440dd1-b0cd-4f23-ac4f-9a2a6acee947) . . . Server cf3 ready (startup time: 15
seconds) There are all sorts of entries about the differenent instances being
discovered, etc. When I changed the cluster name so that the second physical
server was no longer included in the cluster the startup log file looks like:
Starting Macromedia JRun 4 (Build 84683), cf3 server Warning: file url in
codebase component may cause problems (file:/D:/JRun4/lib/jrun.jar!/reggie.jar)
04/06 00:29:32 info ClusterManager: JINI transient lookup service starting
(cluster broadcast may require a moment)... 04/06 00:29:32 info ClusterManager:
JINI transient lookup service started 04/06 00:29:33 info Watching directory,
D:\JRun4\servers\cf3\SERVER-INF\cluster. 04/06 00:29:33 info JRun Naming
Service listening on *:2908 04/06 00:29:34 warning No sessionSecret has been
specified in jrun.xml. Installing a self generated sessionSecret. 04/06
00:29:34 info No JDBC data sources have been configured for this server (see
jrun-resources.xml) 04/06 00:29:34 info JRun Web Server listening on *:8301
04/06 00:29:34 info JRun Proxy Server listening on *:51000 04/06 00:29:34 info
Deploying web application 'file:/D:/JRun4/servers/cf3/SERVER-INF/cluster/'
from: file:/D:/JRun4/servers/cf3/SERVER-INF/cluster/ 04/06 00:29:34 user
JSPServlet: init 04/06 00:29:34 info Deploying enterprise application 'cfusion'
from: file:/D:/JRun4/servers/cf3/cfusion.ear/ 04/06 00:29:35 info Deploying web
application 'cfusion' from: file:/D:/JRun4/servers/cf3/cfusion.ear/ 04/06
00:29:36 INFO License Service: Flex 1.5 CF Edition enabled 04/06 00:29:36 INFO
Starting Flex 1.5 CF Edition 04/06 00:29:36 info Web Services in
cfusion.ear#cfusion.war: 04/06 00:29:36 info (No web services found.) 04/06
00:29:39 warning Could not find valid replication peer- assuming this server is
first server running in replication group. 04/06 00:29:39 user JSPServlet: init
04/06 00:29:39 user CFMxmlServlet: init 04/06 00:29:39 user CFMxmlServlet:
Macromedia Flex Build: 87315.134646 04/06 00:29:39 INFO Macromedia Flex Build:
87315.134646 04/06 00:29:40 user ColdFusionStartUpServlet: init 04/06 00:29:40
user ColdFusionStartUpServlet: ColdFusion MX: Starting application services
04/06 00:29:40 user ColdFusionStartUpServlet: ColdFusion MX: VM version =
1.4.2_05-b04 . . . Server cf3 ready (startup time: 18 seconds) No more mention
of any of the other instances that should be a part of the cluster. It's
almost like the JINI lookup service can't see the machine it's running on, it
can only see other physical servers.

[ke4pym]

My observation has been that once you're assigned to an instance, you'll stay
on that instance and not flip-flop. I verify proper cluster operation by
generating a token and then stopping the instance I'm on and rerunning the
token generator. If all is working, I keep my token and see the instance name
change.

I assume that your cf2 instance is started?

Check
jrun4\servers\<jrun-server-name>\cfusion-ear\cfusion-war\WEB-INF\jrun-web.xml

Are your buddy-names correct (you can also check this in the JRun web admin
console)?