----------------------------------------------------------------
BEFORE YOU POST, search the faq at <http://java.apache.org/faq/>
WHEN YOU POST, include all relevant version numbers, log files,
and configuration files.  Don't make us guess your problem!!!
----------------------------------------------------------------

This one is gonna be a little of a doozy so I'm going to try my best to 
describe what we have :

2 machines
solaris 2.7 on "accustress"
solaris 2.6 on "cpts3"

accustress is a e450
cpts3 is a e250 

accustress ip is : 192.168.1.170
cpts3 ip is      : 192.168.1.171

the jserv.conf on accustress is (load balancing part) :

#ApJServMount /servlets /root
#ApJServMount /servlet /root

ApJServMount /servlet balance://set1/root

ApJServBalance set1 accustress
#ApJServBalance set1 cpts3

ApJServHost accustress ajpv12://192.168.1.170:8070
ApJServHost cpts3 ajpv12://192.168.1.171:8071

ApJServRoute JS1 accustress
ApJServRoute JS2 cpts3

ApJServShmFile /FSERVER/education/lb/jserv_shm

the jserv.conf on cpts3 is (the load balancing part) :

#ApJServMount /servlets /root
#ApJServMount /servlet /root

ApJServMount /servlet balance://set1/root

#ApJServBalance set1 accustress
ApJServBalance set1 cpts3

ApJServHost accustress ajpv12://192.168.1.170:8070
ApJServHost cpts3 ajpv12://192.168.1.171:8071

ApJServRoute JS1 accustress
ApJServRoute JS2 cpts3

ApJServShmFile /FSERVER/education/lb/jserv_shm

There is also apache web server instances on each box... there is an 
ipivot (hardward load balancing device) in front of the two apache web 
servers... this provides the vip that binds back to the .170 and .171 
web servers instances...

Current issues / situations :

The ipivot works great, you hit the vip, and it puts you on either web 
server instance is the least busy.... i can see it flipping boxes in the 
web server logs...

When we hit a servlet, i can see it also shooting us to different jserv 
instances.. we have a "session spy servlet" which allows us to look at 
the corresponding sessions and i see .js1 and .js2

the problem starts when we kill (by hand) one of the jserv instances...

if we kill say accustress instance of jserv (on .170), and we had a 
session on .js1, sometimes we get flipped to .js2 and all is well ... 
BUT.. if you reload again, more times than not, it seems that we get 
thrown on accustress web server instance and instead of routing us to 
.js2, it trys to hit .js1 and of course we get a server error because 
the jserv instance is not up...

is there something wrong in our config?

also, we have the shm file on shared storage right now... im not sure if 
this is correct or not but we have tried both individual files (on each 
machine) and shared storage (ie they both can see the shm file) but 
neither work.. 

we have also tried uncommenting the cpts3 and accustress comments on 
each jserv.conf with no effect, but as we understand from the load 
balancing doc, it seems like the way i have it listed now is correct... 
you balance across the machine you are on and route to the oher... the 
words i used there may not be quite on, but that's the best way i could 
put it ;)

if we go to 192.168.1.170/jserv/, we do see accustress reporting that 
its jserv session as down, but again, the routing does not seem to 
happen?

is there something fundemental going on such as we can not have multiple 
 apache instances across multiple machines?  if so, how can you have a 
totally fault tolerant jserv app?

if anyone needs any more info, or if this info sounds crazy, ask away!

*phew*

-ed


--
--------------------------------------------------------------
Please read the FAQ! <http://java.apache.org/faq/>
To subscribe:        [EMAIL PROTECTED]
To unsubscribe:      [EMAIL PROTECTED]
Archives and Other:  <http://java.apache.org/main/mail.html>
Problems?:           [EMAIL PROTECTED]

Reply via email to