SAM Linux agent issues?

jim.trucanoharp

My team is having some major issues with the Linux agent!

What we have found is that if the Linux agent loses connection to the Orion poller, the linux agent goes into "local" mode and begins writing the polled data to /tmp.

The agent eventually fills up /tmp and crashes the server and/or apps on the server.

There is a SW open case on this currently.

We have two requests in the case to address this:

     1) create a way to limit the amount of data the agent can write to /tmp

     and/or

     2) allow the end user to configure the folder the agent writes to to something other than /tmp so that the Linux admins can manage the amount of data written

Please Thwackers,

If you have similar issues, respond to this so we can provide more input for the SW dev team to fix this.

cheers,

Jim

Senior Engineer

Tobias International

garycheung

I submitted a case with SW last week, and here is their response.

"You can change the cfg to reduce the number of TaskProperties files:

After the connection to the agent goes down, data is being collected and saved. Periodically it gets resent. After a period of time the data will eventually time out and be thrown away. You may set the jobengine configuration parameter jobresult_ttl_extension_mins to 60 or something less than 1440. This means only 60 minutes of data will be saved until connection is reestablished to the agent. This will limit the amount of data resent and should reduce the amount of task files.

The job engine configuration file is /opt/SolarWinds/Agent/bin/Plugins/JobEngine/SolarWinds.Agent.JobEngine.Plugin.cfg"

During the weekend. All the Linux Agents start to report high response time and high packet Loss. I change the agent Status & Response time from "Status & Response Time Agent" to ICMP. They are all good now.

Hope this help!

Gary