2 Replies Latest reply on Feb 10, 2009 6:32 AM by ecnaton

    lots of TIME_WAIT


      I was monitoring ePO with the perfmon plugin ePO installs and was seeing the amount of connections reaching the maximum set in apache.conf (254). To counter this I increased the number of threads to 1000 and increased the time between agent and server communication as well as the randomization factor in new installs. Also added additional repositories to make the load go down. The load as expected went down.

      However the number of connections in a TIME_WAIT state still gradually build up each week and soon I will have to restart apache again to make them go away. Is this a known problem?

      This is not like ePO stops working or serving clients, but is does begin to drop clients after a while. With the message server is busy. Once I restart the apache the number of clients drops to 5-10 because that is the amount of real clients that are connected.

      Has anyone seen this before?