Customer observing below logs very often.

show log messages | match memory | last

May 7 12:32:27 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2524

May 7 12:34:42 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2525

May 7 12:36:32 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2526

May 7 13:16:46 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2527

May 7 13:21:11 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2528

May 7 13:39:18 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2529

May 7 13:43:39 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2530

Customer observing below logs very often.

show log messages | match memory | last

May 7 12:32:27 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2524

May 7 12:34:42 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2525

May 7 12:36:32 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2526

May 7 13:16:46 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2527

May 7 13:21:11 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2528

May 7 13:39:18 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2529

May 7 13:43:39 2024 QFX /kernel: Percentage memory available(19)less than threshold(20 %)- 2530

The low memory seen in this device is expected and not an issue when multiple operations are initiated on the device.

Also, based on applications, this memory goes low, since there is no functionality breakage, this is not an issue.

Details of analysis is below:

1. The system has around total of 2Gb of memory for use.
2. Have taken the memory dump from system, there are instances of free memory % low for this device.(e.g single console connection, with cli command requests on the device is causing multiple daemons getting created to fetch the data from PFE to RE e.g mgd daemon is taking 2% of memory space and multiple instances are getting created based on the requests).
3. Since the system is having low memory and due to multiple mgd instances are getting created in the system, the system is going for low memory.
4. Each request to mgd component from the above layers is taking around 61K, and is consuming the available memory.
5. Have taken the memory dump from system after zeroize the system(in lab) and and seen the memory in normal non traffic scenario also, there are instances of free memory % low for this device.