7.0 Agents May Suffer Random Disconnects (4210006)

Two known issues exist with the new Systems Management 7.0 agent architecture that may cause random disconnects.

• K1-19139: TCP connections between the agent and server can be randomly disconnected. Indicators of this issue - aside from behavior - are only apparent in logs that require backend access (Support Only).
• K1-19104: MySQL timeouts/restarts on the Systems Management server can cause stale agent connections to remain in the server database. Indicators of this issue include the random disconnect behavior and "MySQL has gone away" errors in the "Agent Messaging Protocol v2 Queue" Log (visible in the System UI under Settings | Logs).

In both cases, agents will self-heal using the Watchdog service for each instance when this occurs (agents will not permanently self-heal and must do so each time the issue occurs). The Watchdog service checks for missed inventory intervals on agents and will initiate a restart if 3 inventory intervals have been missed. However, this can take up to 6 missed inventory intervals since, for example, it is possible for the 3 interval threshold to be satisfied by an inventory sent 2.9 intervals ago. Take for instance an inventory interval of 2 hours. It may take up to 12 hours for Watchdog to initiate the agent restart due to missed inventory intervals (up to 6). To find the maximum length of time it should take for Watchdog to restart an agent due to missed inventory intervals, simply multiply the inventory interval by 6.

No hotfix is available at this time.

If either defect is suspected, please contact KACE Support and reference this article.