Soumen
Reputable Poster
Dear List,
I am looking for the correct way to draining connection using the AWS ALB.
Here is our setup:
We have 7 different WebLogic Servers each having 5-6 JVM servers. These weblogic servers are load balanced using AWS Application Load Balancer and is using pretty default setup on AWS ALB side.
What we are trying to achieve:
We are in the process of upgrading the WebLogic on these webserver machines and wanted to leverage the connection draining feature of AWS ALB or for that matter any load balancer. Typically you can take a server in the AWS ALB mix out and if you have connection draining enabled the load balancer will wait for all active connections to terminate (until the time it is mentioned in the deregistration delay - say 300 seconds default). After that time the server is taken out of the load balancer. During this time no new connections are sent to the server as well and all connections are routed to other targets.
What we are observing:
We observed through multiple tests that when we set a target server to de-register and AWS shows that it is currently draining the connections, on JDE side the user who is logged into that particular server is immediately logged off. Based on my discussion with my AWS cloud engineer it is the expected nehaviour because for him any new browsing or transaction at the user end after a server is set to deregister is a "new TCP/IP connection" and will be routed to a different server. I did observe though that once I am kicked out and re-login it directs me to a another target server with a different session id.
Questions:
- Is there a graceful way to handle this instead of just pulling the rug out of the users feet? May be I am missing a step some where?
- Can we possibly use the combination of server manager "disable login" and the AWS connection draining somehow? any ideas?
Just looking for any better way to handle this.
I am looking for the correct way to draining connection using the AWS ALB.
Here is our setup:
We have 7 different WebLogic Servers each having 5-6 JVM servers. These weblogic servers are load balanced using AWS Application Load Balancer and is using pretty default setup on AWS ALB side.
What we are trying to achieve:
We are in the process of upgrading the WebLogic on these webserver machines and wanted to leverage the connection draining feature of AWS ALB or for that matter any load balancer. Typically you can take a server in the AWS ALB mix out and if you have connection draining enabled the load balancer will wait for all active connections to terminate (until the time it is mentioned in the deregistration delay - say 300 seconds default). After that time the server is taken out of the load balancer. During this time no new connections are sent to the server as well and all connections are routed to other targets.
What we are observing:
We observed through multiple tests that when we set a target server to de-register and AWS shows that it is currently draining the connections, on JDE side the user who is logged into that particular server is immediately logged off. Based on my discussion with my AWS cloud engineer it is the expected nehaviour because for him any new browsing or transaction at the user end after a server is set to deregister is a "new TCP/IP connection" and will be routed to a different server. I did observe though that once I am kicked out and re-login it directs me to a another target server with a different session id.
Questions:
- Is there a graceful way to handle this instead of just pulling the rug out of the users feet? May be I am missing a step some where?
- Can we possibly use the combination of server manager "disable login" and the AWS connection draining somehow? any ideas?
Just looking for any better way to handle this.
