From: Pawel Branski (pawel.branski@foreseeresults.com)
Date: Thu Mar 31 2005 - 18:14:58 EST
What may appear to be slightly off topic but I need some help. I did
a lot of search on the net but I was not able to find the answers.
Maybe someone is running similar environment.
We have a couple of web servers in cluster running Solaris 8 and
Solaris 9 with apache and Websphere application server. Websphere
application server is accessing oracle db.
We have two oracle servers in the Oracle Advance Replication
environment, to have an option of directing part of the web cluster
traffic to one oracle server and the second part of the web cluster to
the second oracle server. In case of one of the oracle server failure
we would like to be able to redirect the oracle traffic from the
websphere to oracle server which is still online. If the db crashes,
oracle server can redirect the traffic to the server still online, but
when the server is down due to OS / hardware problems then we have the
problem. Websphere can redirect the traffic but TCP_IP setting on the
apache / application server according to oracle will have to be
changed to make timeout faster.
The following are CLIENT settings suggested that would make NIC
failures on the Server quicker to detect by the client system.
These are OS settings and effect ALL applications, tools and
components running on the System.
tcp_ip_abort_cinterval = 10000 (default is 180000)
tcp_keepalive_interval = 240000 (default is 7200000)
tcp_ip_abort_interval = 60000 (default is 480000)
If I make the changes on the application server (apache / websphere)
what could I break?
Would this effect in any way clients accessing our application?
Any help will be much appreciated.
I will send the summary,
Thank you, pb
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:30:27 EDT