|
Use the following command to check which CPU core is used by each daemon.
The last column shows the CPU core index
diagnose sys top
Run Time: 31 days, 18 hours and 12 minutes
0U, 0N, 0S, 100I, 0WA, 0HI, 0SI, 0ST; 2007T, 663F
cmdbsvr 145 S 4.0 2.7 0
node 28235 S 3.3 2.5 1
updated 194 R 2.1 1.7 4
reportd 190 S 6.0 1.5 3
cw_acd 211 S 6.3 1.4 2
forticron 180 S 2.7 1.3 3
miglogd 189 R 1.1 1.3 1
autod 215 S 5.8 1.2 4
From the above output, it is possible to conclude that the daemon updated process ID is 194 and it is running on CPU core 4.
The following command can also be used to check which CPU core is used by each daemon.
diagnose sys process dump 189 | grep Cpu <----- 189 is the Process ID(PID) of a daemon/process
Cpus_allowed: 2 <----- 0010 (converted by binary).
Cpus_allowed_list: 1 <----- CPU no. 1 (2nd from the far right is '1').
If one core is observed (can be multiple) being used significantly more for a longer time, then, it can be helpful to run the CPU profiler.
get sys performance status
CPU states: 1% user 0% system 0% nice 99% idle 0% iowait 0% irq 35% softirq
CPU0 states: 0% user 0% system 0% nice 100% idle 0% iowait 0% irq 0% softirq
CPU1 states: 0% user 0% system 0% nice 100% idle 0% iowait 0% irq 0% softirq
CPU2 states: 0% user 0% system 0% nice 100% idle 0% iowait 0% irq 0% softirq
CPU3 states: 5% user 0% system 0% nice 95% idle 0% iowait 0% irq 40% softirq
For example: CPU core 3 has the highest load in the system, softirq.
Now, proceed with CPU core profiling, which will give insight into what may be the problem (e.g. offloading).
diagnose sys profile cpumask <cpu core number>
diagnose sys profile start <- Run the command and leave it for 10-15 seconds.
diagnose sys profile stop <- After 10-15 seconds run it to stop the profiler.
diagnose sys profile show <- Will print the result on the console.
diagnose sys profile show order <- Shows in order.
Note:
If the process is in the S state (sleeping state), '# diagnose sys top' will display the last used CPU because the process is sleeping.
Identify the CPU core being used on older versions before FortiOS 7.0:
- Identify the Process ID (PID) of a daemon/process.
diagnose sys top
[JRun Time: 1 days, 23 hours and 54 minutes
0U, 0N, 0S, 100I, 0WA, 0HI, 0SI, 0ST; 7996T, 5392F
ipshelper 330 S < 0.0 1.6
ipsengine 333 S < 0.0 1.2
ipsengine 331 S < 0.0 1.2
ipsengine 332 S < 0.0 1.2
reportd 200 S 0.0 1.1
cmdbsvr 159 S 0.0 0.4
pyfcgid 186 S 0.0 0.4
scanunitd 203 S < 0.0 0.3
scanunitd 239 S < 0.0 0.3
scanunitd 238 S < 0.0 0.3
scanunitd 236 S < 0.0 0.3
scanunitd 237 S < 0.0 0.3
httpsd 185 S 0.0 0.3
pyfcgid 233 S 0.0 0.3
pyfcgid 234 S 0.0 0.3
pyfcgid 235 S 0.0 0.3
httpsd 243 S 0.0 0.3
httpsd 284 S 0.0 0.3
cw_acd 218 S 0.0 0.3
forticron 192 S 0.0 0.2
- After tracing the PID of a daemon, issue the following commands:
diagnose sys process dump 7091 | grep Cpus_allowed
Cpus_allowed: f <----- Hex Value.
Cpus_allowed_list: 0-3 <----- List of CPU cores.
Convert the HEX value to binary to find out the core using the following table:
Decimal Binary CPU used
1 0001 cpu-0
2 0010 cpu-1
3 0011 cpu-0 and cpu-1
4 0100 cpu-2
.......
15 1111 cpu-0, cpu-1, cpu-2 and cpu-3
16 10110 cpu-1, cpu-2, and cpu-4
.......
Use a HEX to Binary converter.
|
