|
Background:
Conserve mode occurs when memory utilization exceeds the configured value under set memory-use-threshold-red 88, typically caused by insufficient memory capacity, unintended software behaviors, or memory-intensive processes. Long-term monitoring of memory consumption trends is essential for identifying memory leaks and preventing conserve mode activation
For more information, see Technical Tip: Memory Management: A Long-Term Strategy to Prevent Conserve Mode
Fortinet identified a memory leak in the NODE on v7.4.8 and v7.6.3 that can be identified through regular monitoring using these commands:
Identifying the Memory Leak Pattern:
Step 1: Monitor process memory consumption.
F2 # diagnose sys top-mem 5
node (2093): 328733kB <----- Increases.
wad (2198): 192759kB
wad (2199): 106893kB
ipsengine (2445): 97662kB
ipsengine (2444): 95252kB
Top-5 memory used: 821299kB
Step 2: Check overall memory utilization.
F2 # get system performance status | grep Mem
Memory: 4041524k total, 2511412k used (62.1%), 1065808k free (26.4%), 464304k freeable (11.5%) ß Total utilization increases
If memory increases consistently over time (several days), a memory leak pattern has likely been identified.
Contributing Factors:
External management tools such as security fabric, FortiManager etc, that query the FortiGate can trigger increased node process activity.
Workaround:
Restart the NODE process. Follow these steps to do so:
- Identify the process ID. This information can be viewed with the following command:
F2 #diagnose sys process pidof node
2093
Or it can be viewed with this command:
F2 # diagnose sys top-mem 5
node (2093) 328733kB
- Restart the process with the following parameter:
F2 # diagnose sys kill 11 2093
This temporarily resolves the memory leak by releasing the memory that was held by the process. Restart the node daemon via an automation script.
config system auto-script
edit restart_node
set interval 86400 <---- 24 hours.
set repeat 1000 <---- 1000 times.
set start auto
set script "fnsysctl killall node" <----- Command to restart the Node.js daemon.
next
end
Fix:
This issue has been resolved in v7.4.9 and v7.6.5 (scheduled for release in November 2025): Resolved issues
Related article:
Technical Tip: High memory usage of node process
|
