Re: AW: Apache 2, Tomcat 5, mod_jk?
Nitschke Michael wrote: mod_jk2 is a rework/reimplementation of mod_jk and is expected to work with Apache 2 and tomcat >4. there are some good tutorials and howtos, for Tomcat >= 4 - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Load Balancing with MOD_JK and MOD_JK2 on windows.
Vladyslav Kosulin wrote: Joe Harmon wrote: Please, please, please, can someone help. MOd_JK2 issue: If I am using mod_jk2.dll (tried several builds) I get the round robin to function, but not the stick sessions. All three instances of Tomcat are running on the same box using the same files with different ports. Here is the configuration. ## This is for the first Tomcat instance ## uses SERVER.XML [channel.socket:lab49.provo.novell.com:9009] port=9009 host=127.0.0.1 Change the names to remove dots everywhere, i.e. use: [channel.socket:lab49:9009] port=9009 host=127.0.0.1 mod_jk2 daes not support dots in names. it is a known bug for a long time. I meant, mod_jk2 support for sticky sessions is broken for names with dots. it can not correctly identify the target host if the JSESSIONID is in form of xxx.hostname.domainname. It can handle only xxx.hostname Vlad - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: MOD_JK2 load balancing example
Joe Harmon wrote:
Is there anyone who has load balancing working with MOD_JK2? If you do
could you please send me a working example of the workers2.properties
file. Also, please let me know whether or not you are running tomcat on
three separate servers, or off of the same box using different
server.xml files.
I use jetty, but anyway here is linux Apache 2.0.47 file:
[logger]
level=INFO
debug=10
[logger.file:0]
file=${serverRoot}/logs/jk2.log
level=DEBUG
debug=10
version=0
[config:]
file=${serverRoot}/conf/workers2.properties
debug=10
debugEnv=10
[uriMap:]
info=Maps the requests. Options: debug
debug=10
[shm:]
info=Scoreboard. Required for reconfiguration and status with multiprocess servers
file=${serverRoot}/logs/jk2.shm
size=1048576
debug=10
[workerEnv:]
info=Global server options
logger=logger.file:0
debug=10
[lb:lb]
info=Default load balancer.
timeout=3
debug=10
[channel.socket:www2:8009]
info=Ajp13 forwarding over socket to www2
lb_factor=1
group=lb
debug=10
version=0
[channel.socket:www4:8009]
info=Ajp13 forwarding over socket to www4
lb_factor=1
group=lb
debug=10
version=0
[status:status]
info=Status worker, displays runtime informations
version=0
[uri:/jkstatus]
info=Display status information and checks the config file for changes.
group=status:
version=0
[uri:/taxgard/*]
#context=/taxgard
group=lb:lb
debug=10
version=0
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
Re: Load Balancing with MOD_JK and MOD_JK2 on windows.
Joe Harmon wrote: Please, please, please, can someone help. MOd_JK2 issue: If I am using mod_jk2.dll (tried several builds) I get the round robin to function, but not the stick sessions. All three instances of Tomcat are running on the same box using the same files with different ports. Here is the configuration. ## This is for the first Tomcat instance ## uses SERVER.XML [channel.socket:lab49.provo.novell.com:9009] port=9009 host=127.0.0.1 Change the names to remove dots everywhere, i.e. use: [channel.socket:lab49:9009] port=9009 host=127.0.0.1 mod_jk2 daes not support dots in names. it is a known bug for a long time. ## This is for the second Tomcat instance ## uses SERVER2.XML [channel.socket:lab49.provo.novell.com:9019] port=9019 host=127.0.0.1 ## This is for the third Tomcat instance ## uses SERVER3.XML [channel.socket:lab49.provo.novell.com:9029] port=9029 host=127.0.0.1 ## This is defining the worker over the AJP13 protocol. This will need to be done for each instance. ## [ajp13:lab49.provo.novell.com:9009] [ajp13:lab49.provo.novell.com:9019] [ajp13:lab49.provo.novell.com:9029] [lb:lb] worker=ajp13:lab49.provo.novell.com:9009 worker=ajp13:lab49.provo.novell.com:9019 worker=ajp13:lab49.provo.novell.com:9029 #tickySession=1 [shm:] info=Scoreboard. Required for reconfiguration and status with multiprocess servers file=c:/apache/Apache2/conf/xx/jk2.shm size=100 debug=0 disabled=0 [uri:/examples/*] worker=lb:lb worker=ajp13:localhost9009 [status:status] [uri:/jkstatus/*] worker=status:status [uri:/admin/*] worker=ajp13:lab49.provo.novell.com:9009 [uri:/nps/*] group=lb stickySession=1 ###End of workers2.properties## Joe Harmon Web Services Support Novell Inc. - The leading provider of Information Solutions. (801) 861-9163 [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Urgent !!! Problem to get TOMCAT/4.1.24 cluster running withsticky session support.
[EMAIL PROTECTED] wrote: Hi , some additional information . I've got with mozilla LiveHTTPHeaders that my session cookie doesn't recieve Jvmroute parameter , so apache doesn't know anything about load-balancing . Cookie: JSESSIONID=5387242C819757A9BC12B2FAF1AF2AD8; Does anybody have any suggestion or idea? Is your web application ? I don't know how Tomcat handles this, but Jetty does not attach jvmRoute for distributable applications assuming that sticky sessions are meaningless in this case. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Urgent !!! Problem to get TOMCAT/4.1.24 cluster running withsticky session support.
[EMAIL PROTECTED] wrote: Hi guys here is detail description: I have Apache/2.0.45 running on the server A If your Apache is running on UNIX/Linux/BSD/MacOS X, load balancing with sticky sessions will work only if you use worker MPM. At least this is correct for mod_jk2, and may be the same is the cause for mod_jk. By default Apache is been compiled with prefork MPM on UNIX/Linux/BSD. Check with httpd -l If you see prefork.c, you have to recompile Apache using ./configure --with-mpm=worker ... Hope this will help. Vlad And two tomcat workers are running on B and C The problem is that I cannot get Tomcat cluster load balanced , playing around with workers.properties on Apache gave me two different situations. 1.if I have local_worker parameter equal to 1, then I have no lost sessions but also no loadbalancing . But cluster is still fail safe, if one tomcat dies - another one gets all incoming requests. 2. if I have local_worker=0 then I have a simple round- robin balancer without session affinity. So my session got lost. configuration example workers.properties: worker.list=tomcat-worker-01,tomcat-worker-02,router worker.tomcat-worker-02.port=4007 worker.tomcat-worker-02.host=xxx.xx.x.12 worker.tomcat-worker-02.type=ajp13 worker.tomcat-worker-02.lbfactor=50 worker.tomcat-worker-02.cachesize=10 worker.tomcat-worker-02.cache_timeout=600 worker.tomcat-worker-02.socket_timeout=300 worker.tomcat-worker-02.local_worker=1 worker.tomcat-worker-01.port=5007 worker.tomcat-worker-01.host=xxx.xx.x.20 worker.tomcat-worker-01.type=ajp13 worker.tomcat-worker-01.lbfactor=50 worker.tomcat-worker-01.cachesize=10 worker.tomcat-worker-01.cache_timeout=600 worker.tomcat-worker-01.socket_timeout=300 worker.tomcat-worker-01.local_worker=1 worker.router.type=lb worker.router.balanced_workers=tomcat-worker-01,tomcat-worker-02 worker.router.sticky_session =1 worker.router.local_worker_only=1 P.S. I checked the previous discussions related to server.xmlS configuration. I had there: -- Yefym Dmukh developer email: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
mod_jk2 load balancing as a RAID 0+1 analog
Hi all, Here is my configuration: Front-ent - Apache 2.0.47/mod_jk2/ajp13. 4 instances of JBoss 3.2.1/Jetty 4.2.11 with identical ear applications are configured as 2 pairs of fail-over clusters with webapp. node1+node2 creates a first fail-over cluster. node3+node4 creates a second fail-over cluster. Initial http request can go to any of these 4 nodes. After that mod_jk2 should redirect subsequent requests inside the initial fail-over cluster, and only if both nodes in that cluster fail, try to redirect to another nodes. The problem here is that I do not know is it possible to configure mod_jk2 (workers2.properties) to redirect this way. Any ideas are highly appreciated. Thanks, Vlad - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
