Pi-Star 4.0 rc2 CPU load higher, temps up

[N6WBL]

Anyone running v4.0 rc2 and seeing higher CPU load and temps? I reverted to 3.4.17 on my DVMega for temps and 3.4.17 on my JumboSpot while I toyed with upgrading the right firmware. Both devices were registering higher CPU loads and temps with v4.0 than v3.4.17 is showing. This load just the new base OS and kernel?

Jason - N6WBL

[VK7HSE]

I noticed that as I have two hotspots that are running YSF and they both try to report into my tier2 service whatever one gets in first will cause the other one to get it's log flooded with the tier2 server advising that in my case VK7HSE-ND is already logged in! So when this happens I've noticed that grep sits on anything between 60-95% cpu. At this stage the the offender is grep being used by the www-data group.

Might not be related to the above mentioned issue, to remedy this issue I have to change the ssid of the one that raspberry pi's and it comes good

Sent from my Pixel 2 XL using Tapatalk

[N4NQY]

I have been running 4.0 for days 24/7 with MMDVM_HS Hat on a Pi 3B (not Plus), running WiFI.

CPU load (DMR only) is 1.15% and the Temperature is 48.9 C or 120 F.

I enabled YSF and did not notice a change after 15 minutes.

This is about the same as 3.1.16/17 was for me.

I have a Pi 3B+ in another hotspot and it runs 0.05 CPU load and 55.8 C or 132.4 F. The B+ runs a lot hotter than the 3B even with heat sinks, which I think is required for both versions of the Pi 3.

I am not sure I know why you are seeing higher CPU loads. "if all else fails, reboot"

N4NQY

[W0RMT]

I've definitely seen higher loads and higher temps on at least one of my hotspots: pi 3B+ with DB9MAT et al duplex board. It now routinely runs >120F when in use, and previously it never ran above ~114F. I have not investigated processes, etc at this point but have only observed the increased T.

[N6WBL]

Upon playing around with SSH and pistar-update, looks like the SSH command first pulls OS updates before disabling the hotspot and applying Pi-Star updates unlike using the web update which doesn't pull OS updates and jumps right into the Pi-Star updates. So I tried both my DVMega/Pi 3 and Jumbo hotspots with v4.0 rc2 again. After applying the OS updates via SSH I've seen my load levels and temps return to normal. The thing that jumped out at me running top was maybe it was a systemd related as there were updates for systemd. So there is a reason to SSH in and run pistar-update to apply OS updates in addition to the Pi-Star updates going through the web menu.

For fun I ran the updated v4rc2 along an updated v3.4.17 with ttyload to graph loads on both with my DVMega/Pi3 for comparison and they're pretty similar now.
v3.4.17


v4rc2


Jason - N6WBL

[W0RMT]

I've just discovered another issue with my pi-3B+ / DB9MAT et al duplex board on DMR with 4.0 RC2: not only is it running hotter (even after CLI update), but it's dropping me on DMR after ~60 s during each transmission. I have never noticed that behavior before when on 3.4.16 with the same configuration and radio (Anytone 868). In fact, on the previous version it worked very reliably on DMR.

It is currently running at 50C. Previously w/3.4.XX it would run at 42-43C under the same operating conditions.

[N6WBL]

Well, after a few hours the v4.0rc2 load went up again. The ttyload graphs were run without glances active. Then glances was run for a screenshot and includes uptime. Nothing stood out to me, but maybe I missed something. The extra load works out to be about an increase of 4-5C pushing my unit into the orange.

v4.0rc2:


v3.4.17:


v4.0rc2:


v3.4.17:

[K0TI]

I've been doing some testing of this and can replicate it. It is caused by the MMDVN log file getting bigger over time. What I'm seeing is that as the log file gets bigger the 'grep' process that is being used to update the dashboard takes more and more process time. If you make a really large file (10meg) in my test case you end up with 2 or 3 grep process's that run and 100 percent for many seconds at a time, with that continually repeating. In that case the CPU load (using top) went up to almost 3, which I believe means that 3 of the 4 cores where near 100 percent.
Version 3 doesn't act like this. The grep process seems to not take near as much process time to run on Ver 3. Even with the large log file the grep process sometimes will show up at 20-30 percent but only for short periods. With normal log file sizes you rarely see the grep process show up in the top list. If you want to see the effect just put your the hotspot on TAC310 for a while and your average load will be soon creep up over 1 along with the heat rise.

This is running a DVMEGA (single radio) and Pi 3B.

Hope this helps.

Dan K0TI

[N3MNT]

Running 4.0.0-RC2 on Zumspot with Pi0. Usually temp runs 94F when running ver 3.4.16. Same conditions under 4.0.0-RC2 it is running 105F.

[KE0FHS]

My experience matches with what Dan, K0TI is reporting. A couple evenings ago, I was linked to one talkgroup for two hours. the first hour I was just listening to occasional chatter; during the second hour, there was a Hotspot Discussion net that I participated in. During the second hour, the temps on my hotspot (ZUMspot + RPi 3B+ with heat sinks in a well-ventilated case) went up to 59 C, whereas in 3.4.x, that hotspot would max out at 50 C (and I thought 50 C was a bit warm!).