found the problem - indeed it was the firewall. It was blocked off.
But since the connections are stateful and had been established, they
hung on until I restarted the agents. As each agent was restarted, the
firewall block for the return acknowledgment was blocked and thus the
problem of the missi
Actually I was wrong - it was NOT being blocked - but it exhibited
what looked like blocking -- it was the IP tuning params and max
clients causing a backup of the queue and the server, which was not
overloaded, could not clear the queue fast enough. Increased max
agents to 2048 and bumped several
For the archives:
How many agents did you have on that server?
Which "IP tuning params" did you bump?
On Tue, Apr 26, 2011 at 11:49 AM, Kat wrote:
> Actually I was wrong - it was NOT being blocked - but it exhibited
> what looked like blocking -- it was the IP tuning params and max
> clients caus