DynaTrace Managed randomly drops agents

[bbaratono]

We have about 30 oneagents attached to our Server, it should be noted that they are Weblogic on Solaris so they are manually called as autodiscovery does not support Solaris. We have noticed that the Server will randomly lose connection to random agents and the connects will randomly come back. If I reboot my services they will usually all reconnect for a time before random ones start dropping and reconnecting.

Anyone have thoughts on this?

[Julius_Loman · Accepted solution]

Did you check the agent logs? I believe this is something on the network level playing with your connections. Maybe firewall closing connections? I've seen environments where firewalls were closing TCP connections after just few minutes of inactivity.

[bbaratono]

I'm checking with my Network Engineer on that, he says timeout is an hour though, so if it turns out to be the timeout, I guess the question is, can the agent be set to send keep alives?

[Julius_Loman]

Normally agents are "pushing data" to activegate or cluster regularly, so there definitely should some data be sent in a one hour long time slot. Maybe some rate limiting at firewall/proxy? Speaking of proxy - do your agents use proxies?
Anyway, be sure to check the agent logs first.