All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
C2750 Stuck at Base Frequency
Hello,
I have a dedi with an Intel C2750 processor running Ubuntu 18.04.1 LTS (kernel 4.15.0-34).
It's running a quite intensive data-analysis web app, and sometimes load reaches 4+. However, it's always running at 1.2Ghz according to cat /proc/cpuinfo | grep "MHz"
and > lscpu | grep "MHz".
When I run cpufreq-info
, I get
driver: acpi-cpufreq
CPUs which run at the same hardware frequency: 7
CPUs which need to have their frequency coordinated by software: 7
maximum transition latency: 10.0 us.
hardware limits: 1.20 GHz - 2.40 GHz
available frequency steps: 2.40 GHz, 2.40 GHz, 2.30 GHz, 2.20 GHz, 2.10 GHz, 2.00 GHz, 1.90 GHz, 1.80 GHz, 1.70 GHz, 1.60 GHz, 1.50 GHz, 1.40 GHz, 1.30 GHz, 1.20 GHz
available cpufreq governors: conservative, ondemand, userspace, powersave, performance, schedutil
current policy: frequency should be within 1.20 GHz and 2.40 GHz.
The governor "performance" may decide which speed to use
within this range.
current CPU frequency is 1.20 GHz.
cpufreq stats: 2.40 GHz:5.80%, 2.40 GHz:0.00%, 2.30 GHz:0.00%, 2.20 GHz:0.00%, 2.10 GHz:0.01%, 2.00 GHz:0.00%, 1.90 GHz:0.01%, 1.80 GHz:0.03%, 1.70 GHz:0.02%, 1.60 GHz:0.02%, 1.50 GHz:0.05%, 1.40 GHz:0.55%, 1.30 GHz:0.82%, 1.20 GHz:92.68% (717082)
So this output seems to indicate that speed scaling to 2.4Ghz is working sometimes.
However, if I run something like stress --cpu 8
, the CPUs are still at 1.2Ghz even though load is 10+.
I thought maybe it could be a thermal issue, but it seems not:
coretemp-isa-0000
Adapter: ISA adapter
Core 0: +25.0°C (high = +98.0°C, crit = +98.0°C)
Core 1: +25.0°C (high = +98.0°C, crit = +98.0°C)
Core 2: +26.0°C (high = +98.0°C, crit = +98.0°C)
Core 3: +26.0°C (high = +98.0°C, crit = +98.0°C)
Core 4: +24.0°C (high = +98.0°C, crit = +98.0°C)
Core 5: +24.0°C (high = +98.0°C, crit = +98.0°C)
Core 6: +25.0°C (high = +98.0°C, crit = +98.0°C)
Core 7: +25.0°C (high = +98.0°C, crit = +98.0°C)
So I wondered if anyone on LET had any suggestions please
Comments
Online?
Yeah (well, OneProvider).
On my Dedibox with a C2750, when I run stress, the cores jump to 2400 MHz.
Edit: Centos 7
Thanks - it's useful to know it's not the standard behaviour.
Mine definitely don't jump - here's htop, stress and the CPU Mhz.
I know the issue from a few people, I also know, that OneProvider sells Online crap hardware, thats why its so cheap.
A "permanent" fix would be, reboot the server daily.
Seems to be a hardware issue, as far as I know.
Well, I rebooted the server and now it's running at 2.4Ghz under load. That was underwhelming
Ah so it seems like rebooting it is known to be the fix. Thanks I know Online aren't always amazing (especially with discount pricing nowadays), but I've been with OneProvider for a long time now and they've treated me well.
Have you opened a ticket?
You could try something like this in your kernel commandline:
cpuidle.off=1 intel_pstate=disable intel_idle.max_cstate=0 nohalt idle=poll
This should disable all cpu power management/c-states, etc.
Thanks man, that's a great suggestion I did that after the reboot and it all seems to be working properly now.
And Online will hate him for that.
I guess their point is that power/electricity and cooling are a major cost factor, so they strongly prefer (trying to put that politely) the machines in power save mode.
It's a dedicated machine.
Either I am going to do what I want with it or: OK, thx, bye.
Simple.
Oh, I certainly fully agree!
I merely tried to shine some light on the "why", why their dedis seem to not only in rare exceptional cases exhibit that ugly attribute.
Could be some custom solution at boot up.
But that can be identified.
I am using this when i need to put the walls to the balls:
And this, to return it back to ondemand.
I love the fact that this post was blissfully ignored.
I didn’t respond to it because these are unmanaged services and issues with CPU scaling are almost always user configuration issues or software bugs (as it was in my case).
That being said, it would have been polite of me to answer, but I’m quite busy at work and it slipped my mind.
Thanks, that’s a good suggestion! I actually already had the Performance Governer enabled (it shows in my main post), but it wasn’t working until I rebooted