Agents may compete with other applications for limited Random Access Memory (RAM).
Note: In addition to the parent process, we monitor all child processes, so they all can be throttled.
Through the JetPatch Agent Manager, you can limit the amount of Random Access Memory that each managed agent is allowed to consume.
This can be done for all the managed agents under the Intigua connector and the Managed Agents by navigating to the “Tools Catalog” under “Agents & tools” on the left side of the interface
To limit total RAM consumption of all the agents managed under a specific connector, Select the Connector Icon and navigate to the “Services” tab. Select the connector you wish to manage and navigate to the section entitled “CPU and Memory Control”
It should resemble this image:
To limit total RAM consumption of a specific agent managed under a specific connector, you must choose a specific Agent rather than the Intigua connector in the “Tools Catalog” section, the same navigation applies to reach the “CPU and Memory Control” section.
It should resemble this image:
The limit placed on the Connector will supersede all the limits governing the Managed agent Ram consumption.
Take a scenario where the Intigua connector's memory consumption is limited to 512 MB. Even if the limit governing the managed agent is higher than that, the most RAM that managed agent can ever use is 512MB
--------------------------------------------------------
For each endpoint, all of the managed agents will be throttled equally in order to meet the configured endpoint RAM consumption threshold.
Take a scenario where Intigua connector RAM limit: 512MB
The Limits configured for the managed agents are as follows:
- Agent1 is 256MB
- Agent2 is 256MB
- Agent3 is 256MB
- Agent4 is 256MB
If in real-time, the Ram consumption of the managed agents is:
- Agent1 is using 512MB
- Agent2 is using 512MB
- Agent3 is using 512MB
- Agent4 is using 512MB
Then they will be throttled equally to their limits at 256MB.
Related Articles:
Comments
0 comments
Please sign in to leave a comment.