Forum Discussion
Chris_Cohee_101
Nimbostratus
NimbostratusLB::reselect Issues
I have the following iRule:
when HTTP_REQUEST {
set respond 0;
HTTP::header insert WL-Proxy-Client-IP [IP::client_addr]
if { [HTTP::uri] starts_with "/channels/"}{
set backup_pool pool_Channels
pool pool_Channels;
}
elseif { [HTTP::uri] starts_with "/autoItemCreate/"}{
set backup_pool pool_NSOProxyServlet
pool pool_NSOProxyServlet
} else {
set uri_default [HTTP::uri]
log local0. "Hit the default rules and the URL is $uri_default";
set uri_new "/portal$uri_default"
log local0. "Hit the default rules and the NEW URL is $uri_new";
HTTP::uri $uri_new;
set backup_pool pool_ProxyToPortal
use pool pool_ProxyToPortal
}
}
when LB_FAILED {
log local0. "Hit the Load Balance Failed rule. The backup pool is $backup_pool";
LB::reselect $backup_pool
}
We are trying to get the pool re-selection to work, but still seeing failures. Any suggestions? Is this possible?
Thank you,
- Chris
hoolioCirrostratus
Hi,
What kind of failures are you seeing? Are all of the nodes in the pools being marked down by monitors? Do you see any relevant log events in /var/log/ltm when a failure occurs?
Aaron
Brian_Curnow_87I'm the WebLogic admin working with Chris on this issue. We are load balancing between nodes of a WebLogic cluster (there are two nodes). When we take a node down there is a period where the load balancer continues to forward requests to the down node (the period is equal to the interval in the monitor).
What we'd like to do is detect the down node in the iRule and select the other node in the cluster. If both nodes are down then it would be appropriate for the load balancer to return an error.
I thought from the docs that if the server were completely down we'd hit the LB_FAILED event and be able to reselect a node from the pool. We're not seeing that happen.