Hi.
Hawing two OPNsense installations. The CPU´s peek every 10 seconds. It was on 25.1 and the same on 25.7.
How high is "high"? The graph looks like "4", which doesn't seem high to me. I couldn't get a periodic blip on mine, and "Reporting: Health" -> "System, processor" doesn't seem to store with sufficient granularity for anything to show up (flat line). Think of a 10 second process - a pinger (monitor), perhaps?
cpu load stats from top (also a widget) is all that really matters.
Seeing a graph that says cpu is doing something, is not all that interesting.
Quote from: BrandyWine on July 30, 2025, 01:55:36 AMcpu load stats from top (also a widget) is all that really matters.
Seeing a graph that says cpu is doing something, is not all that interesting.
Ah, but top isn't so good for looking for patterns or historical data, so I start with the graphs. But your point is good.
(I didn't even notice that the CPU widget always shows a grid height of four, since my CPU almost never gets that high.)
What do the grid line numbers mean? A nice "uniform" bell curve in that widget, that alone seems odd because there's nothing the CPU is doing that looks like that. It usually just ramps up almost instantaneously, and then dwindles down at a fast pace, so I would expect to see more of a sawtooth pattern.
I don't know, CPU widget doesn't see all that useful to me.
That whole "Lobby" is more for impressing "decision makers" and about "competitiveness" with comparators ;-)
Spinned up an OPNsense test instance. Same issue there.
It looks to me like the process causing these regular CPU bumps is "configd.py console".
Quote from: demyers on August 02, 2025, 08:36:50 PMIt looks to me like the process causing these regular CPU bumps is "configd.py console".
Or maybe /usr/local/opnsense/scripts/netflow/flowd_aggregate.py
What does the scale numbers mean? Is it supposed to be total % of all CPU?
The only bumps I get is from Suricata, which has an 11G image, and on my N150 does a bounce bump from about 1-7% for very short time.
It is significant.
Quote from: BrandyWine on August 04, 2025, 09:44:01 PMWhat does the scale numbers mean? Is it supposed to be total % of all CPU? [...]
It appears to be "utilization/number of threads". I can spike one of eight cores (8C/16T CPU) with a log search, which indicates 7% utilization (~1/16) - a bit misleading, as the actual utilization is more like 13%. 8 busy (100%) threads would probably read 50%, while 9 would read 57% while getting little more work done. Not an issue for the N150, of course.
Quote from: robert.haugen@gmail.com on August 19, 2025, 05:31:49 PMIt is significant.
Any spike on a 2C/2T virtual CPU will be more noticeable than with more threads, but I don't see any sort of regular spike on mine. Lots of configuration possibilities for that.
The scale is a bit interesting (not diagnostic). 2C/2T for both: 4.2GHz Coffee Lake @ 15%, vs. 3.9GHz Twin Lake @ 30%. Power saving for the E cores?
I'm seeing this too (GMKtec G2 plus, N150). Every 10 seconds, like clockwork. Doesn't truly concern me except that I have had some instability. Right now I have a cron job to reboot every 24 hours, which seems to work to keep me from going unstable but it's not great. Don't know if related.