Hi,
We are using MarkLogic XCC api from our java code to make calls to MarkLogic XDBC server and both of our Java servers and ML servers run on AWS EC2 instances. We have an ELB above our MarkLogic E-nodes and the IP addresses of this ELB changes frequently but it looks like our code does not seem to take that into effect and is still trying to make calls to old IP of the Elastic load balancer, thus resulting in connection time out issues whenever an ELB IP address changes. While this did not cause any issues with the java client code that connects to ML http servers, it seems to be issue with XDBC servers. I am not sure whether the apache http client is refreshing the connection ip addresses frequently and hence it is working fine but the XCC api is not doing that and thus resulting into issue or there is something else causing this problem. We tried following options to solve our issue. 1. Overrode the networkaddress.cache.ttl in java.security file in the jre/lib/security folder and set the value to 10 seconds. 2. Passed the argument -Dnetworkaddress.cache.ttl=10 to our server startup script. Both options tried separately and together showed no improvement. If any of you already experienced this issue and solved it, can you please let me know what I can do to fix this? Thanks. Note : Since we could not force the ELB to change the IP numbers for the DNS names, we changed the /etc/hosts file on the EC2 that runs our Java client server to have an entry for the ML E-node group ELB DNS name and kept changing the IP to test it. We are using 8.0-6.4 version of ML and our XCC version is of ML7 (I know we need to upgrade this to be in sync with ML version and we will be doing that soon) -Sudheer
_______________________________________________ General mailing list General@developer.marklogic.com Manage your subscription at: http://developer.marklogic.com/mailman/listinfo/general
