Plugin randomly dies with a connection shutdown error message

[jtlisi]

time="2016-12-13T19:31:31-05:00" level=error msg="collector run error" _module=scheduler-job block=run error="connection is shut down" job-type=collector

Snap daemon version (use snapteld -v):
snapteld version test-c2aef10

Environment:

• snmp udp endpoint at port 161

Task Config:

      "config": {
        "/intel/snmp": {
          "snmp_agent_name": "*********",
          "snmp_agent_address": "*******:161",
          "snmp_version": "v2c",
          "community": "*******",
          "network": "udp",
          "timeout": 15,
          "retries": 0
        }
      }

I've tried numerous different timeouts and retry configurations and nothing seems to work. One thing that seems to cause is my setfile has snmp elements in the namespace. Also it will work for hours at a time and then suddenly drop. Once the error message is shown the plugin does not recover regardless of the max-failures.

[nanliu]

@jtlisi, please review intelsdi-x/snap#1456, and see if the daemon control setting max_plugin_restarts to a high value resolve this. If you are able to try the latest build, you can set max_plugin_restarts: -1 to disable this behavior. We also provided additional logging message to indicate the plugin has been disabled due to multiple failures:

level=warning msg="plugin disabled due to exceeding restart limit: 10"

A linux test build of the latest version of snap is available at:
https://s3-us-west-2.amazonaws.com/snap.ci.snap-telemetry.io/snap/latest/linux/x86_64/snapteld

[nanliu]

Based on other user's feedback and our test results, we believe this is fixed per comment above. If you still run into this issue, please feel free to reopen this issue.