This is expected to be added to the proxy
module, but in a way which is more inline with
expectations for a specific proxy server (whether
serving ajp or http or anything else).
On Nov 1, 2006, at 5:39 AM, Rainer Jung wrote:
Hi Gary,
from my understanding of the code mod_proxy(_balancer) at the
moment is
not able to do that. There are status settings "disable" and "stopped"
one can set, but at the moment disbled and stopped behave the same. I
didn't really try, but I derive that from looking at the code. This
holds true for version 2.2.3, the head of branch 2.2 and trunk as
well.
I remember some mail on httpd-dev, that this is known and will be
corrected.
The module mod_jk also has the concept of disable and stop. For mod_jk
disable means, that requests with sessions on a disabled target still
get routed there, but requests without a session do not get routed
to a
disabled target. So if your sessions are relatively short, you can
considerably decrease the number of remaining sessions on a target by
disabling it some time before your maintenance.
I'm sure, this behaviour will come to mod_proxy too, but it looks like
it's not there yet. Anybody: please correct me if I'm wrong.
If you are going to use mod_jk, please get the latest release 1.2.19.
It's important to know, that mod_jk can only handle ajp backends.
Regards
Gary Feltham schrieb:
Hi,
I have a load balancing setup using Apache 2.2.2 (Win32) and 2
instances of Tomcat 5.5.15 and 5.5.17. So far I have been able to
establish a successful load balancing using sticky sessions with the
following config:
ProxyPass /test balancer://testcluster stickysession=JSESSIONID
nofailover=On
<Proxy balancer://testcluster>
BalancerMember ajp://127.0.0.1:18009/test smax=5 route=node01
BalancerMember ajp://127.0.0.1:28009/test smax=5 route=node02
</Proxy>
I am not wanting to setup a tomcat cluster for this deployment but
have 2 independant instances of tomcat running independant sessions.
Currently this is also setup to not failover if an instance goes
down.
For instance node01 is shutdown, the status is:
LoadBalancer Status for balancer://testcluster
StickySession Timeout FailoverAttempts Method
JSESSIONID 0 1 byrequests
Worker URL Route RouteRedir Factor Status
ajp://127.0.0.1:18009/test node01 1 Err
ajp://127.0.0.1:28009/test node02 1 Ok
All expected and a request to node 1 results in a 503 response.
My question therefore is:
Within the context of a scheduled down time such as redeployment of
the application or reboot of the server I would like to be able to
control closure of the load balancer. For example, node 1 is targeted
to be shutdown for maintenance, currently I have 10 user sessions on
that instance and say 9 on node 2. If new users join then the load
balancer will evenly distribute to each node per request. What I want
to control is that each NEW user should only go to node 2, then each
session on node 1 will expire eventually leaving 0 sessions on
node 1.
Node 1 can then be removed cleanly for maintenance and its status set
as disabled.
Is there a current method of performing this in either mod_proxy or
mod_jk - I don't mind which but would like to gain that control?
Many thanks,
Gary Feltham
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
