CPU usage through the roof after latest updates!

pchristy · 06-07-2018, 04:19 AM

System: Slackware 64 current + AlienBob's KDE5
Kernel: 4.15.4
Processor: Intel i7

Installed the latest updates this morning - mostly seems to be fonts. Following the upgrades, I noticed that CPU usage has surged. Top reports:

Code:

 
 2921 pchristy  20   0 3924160 177100 101188 S 110.3   1.1   1:10.20 plasmashell                                       
 1085 root      20   0   60240   7896   5740 S  60.0   0.0  30:43.78 cupsd                                             
 3336 pchristy  20   0  198280  24172   9080 S  34.0   0.1   0:20.05 hp-systray                                        
  966 message+  20   0   21060   3436   2144 R  29.3   0.0  16:17.29 dbus-daemon                                       
 1632 lp        20   0   52020   5532   4968 S  25.3   0.0  13:58.10 dbus                                              
 3190 lp        20   0  312060 265700   5128 S  20.0   1.6   0:11.45 dbus                                              
 2955 pchristy  20   0  260904  22924  19124 S   0.7   0.1   0:00.33 gkrellm

Mostly seems to be plasmashell, but others showing high "idle" usage as well.

Code:

  tail /var/log/syslog
Jun  7 09:17:22 Lightning last message repeated 964 times
Jun  7 10:07:41 Lightning console-kit-daemon[969]: WARNING: Failed to let cgmanager know that it can remove the cgroup when it's empty, the error was: invalid request 
Jun  7 10:07:42 Lightning pulseaudio[2946]: [pulseaudio] pid.c: Stale PID file, overwriting.
Jun  7 10:07:42 Lightning pulseaudio[2946]: [pulseaudio] bluez5-util.c: GetManagedObjects() failed: org.freedesktop.DBus.Error.ServiceUnknown: The name org.bluez was not provided by any .service files
Jun  7 10:07:42 Lightning pulseaudio[2991]: [pulseaudio] pid.c: Daemon already running.
Jun  7 10:07:42 Lightning baloo_file: KDE Baloo File Indexer has reached the inotify folder watch limit. File changes will be ignored.
Jun  7 10:07:42 Lightning last message repeated 28 times
Jun  7 10:07:42 Lightning /hp-systray: hp-systray[3022]: error: option -s not recognized
Jun  7 10:07:42 Lightning baloo_file: KDE Baloo File Indexer has reached the inotify folder watch limit. File changes will be ignored.
Jun  7 10:07:42 Lightning last message repeated 850 times

Anyone know what's going on - or how to fix it?

TIA,

--
Pete

gmgf · 06-07-2018, 05:05 AM

It seem that is a cups problem, because stopped cups, solve the problem.

Darth Vader · 06-07-2018, 05:07 AM

At least in my 2 test/playground computers, from mysterious reasons the PlasmaShell's CPU usage reached about 50% of all 4 cores, BUT I noticed also CUPS in the crowd.

How on those particular computers I do not need printing right now, I stopped CUPS at all, then PlasmaShell started to behave too.

Edit: @GMGF beat me with 2 minutes!

pchristy · 06-07-2018, 05:10 AM

You are quite right! Many thanks!

Is this just occurring on KDE5, or is anyone seeing it on a "stock" install as well? ie: Do we flag this up to Eric or Pat? I ask because it looks as if cupsd is causing plasmashell to go into a minor meltdown!

--
Pete

gmgf · 06-07-2018, 06:10 AM

It seem only plasma5 is affected, tested on one other install with kde4, no problem.

Darth Vader · 06-07-2018, 06:35 AM

I confirm that this behavior is not present under KDE4, even in the same computer, but in another partition, where's installed the standard slackware-current.

kjhambrick · 06-07-2018, 06:45 AM

Quote:

Originally Posted by Darth Vader

I confirm that this behavior is not present under KDE4, even in the same computer, but in another partition, where's installed the standard slackware-current.

Ditto.

Slackware64 14.2+current box

All Updates thru Wed Jun 6 22:19:01 UTC 2018 have been installed, running 4.14.48 GENERIC + latest intel-microcode.

Though I am running the box at runlevel 3, cups is running and when I connect to KDE4 on this Remote Box via rdesktop + xorgxrdp, the load remains reasonable.

The Konsole session in the foreground is an SSH Session to the remote machine where I ran top while connecting to KDE4 via rdesktop

HTH

-- kjh

Toutatis · 06-07-2018, 07:37 AM

I had the same problem after the slackware64-current update this morning. Moreover it was also impossible to use the printer.

This seems to come from the new kernel 4.14.48, because everything works now, after using the preceding 4.14.47

pchristy · 06-07-2018, 08:07 AM

Not sure its to do with the kernel - I'm running 4.15.4!

But! Just restarted cupsd to print something, and now CPU usage is back to normal! Perhaps something in the login process to KDE5 is triggering it?

--
Pete

Toutatis · 06-07-2018, 08:32 AM

Quote:

Originally Posted by pchristy

Not sure its to do with the kernel - I'm running 4.15.4!

But! Just restarted cupsd to print something, and now CPU usage is back to normal! Perhaps something in the login process to KDE5 is triggering it?

--
Pete

You are right. I had also downgraded cups to 2.2.7. But after upgrading again to 2.2.8 and restarting cupsd the problem comes back.
So it comes from cups-2.2.8

sombragris · 06-07-2018, 12:44 PM

Thanks to all for the responses, especially for the tip about cups.

I can confirm the issue.

Apparently this issue appears in the cups + Plasma 5 combination. This issue is not present with the latest cups and Willy's MATE Desktop. However, boot into Plasma 5 and CPU usage goes through the roof until cups is stopped. Main culprits appear to be cups, plasmashell (two processes) and hplip.

SqdnGuns · 06-07-2018, 02:01 PM

Maybe related??

I started seeing this after the update last night, did not have it previously.

Code:

11.946400] CPU0: Core temperature above threshold, cpu clock throttled (total events = 1)
[   11.946401] CPU4: Core temperature above threshold, cpu clock throttled (total events = 1)
[   11.946402] CPU5: Package temperature above threshold, cpu clock throttled (total events = 1)
[   11.946403] CPU6: Package temperature above threshold, cpu clock throttled (total events = 1)
[   11.946404] CPU7: Package temperature above threshold, cpu clock throttled (total events = 1)
[   11.946404] CPU2: Package temperature above threshold, cpu clock throttled (total events = 1)
[   11.946405] CPU3: Package temperature above threshold, cpu clock throttled (total events = 1)
[   11.946406] CPU0: Package temperature above threshold, cpu clock throttled (total events = 1)
[   11.946407] CPU1: Package temperature above threshold, cpu clock throttled (total events = 1)
[   11.946407] CPU4: Package temperature above threshold, cpu clock throttled (total events = 1)
[   11.972385] CPU4: Core temperature/speed normal
[   11.972386] CPU6: Package temperature/speed normal
[   11.972387] CPU0: Core temperature/speed normal
[   11.972388] CPU2: Package temperature/speed normal
[   11.972388] CPU0: Package temperature/speed normal
[   11.972389] CPU4: Package temperature/speed normal
[   11.972430] CPU3: Package temperature/speed normal
[   11.972431] CPU1: Package temperature/speed normal
[   11.972432] CPU5: Package temperature/speed normal
[   11.972432] CPU7: Package temperature/speed normal

alex14641 · 06-07-2018, 05:45 PM

Bug report: https://github.com/apple/cups/issues/5325

chrisretusn · 06-07-2018, 10:38 PM

Quote:

Originally Posted by Toutatis

I had the same problem after the slackware64-current update this morning. Moreover it was also impossible to use the printer.

This seems to come from the new kernel 4.14.48, because everything works now, after using the preceding 4.14.47

Booted back to kernel 4.14.47 a few minutes ago, problem still exist, not the kernel it seems.

I reverted back to cups 2.2.7 and the problem has gone away. Many thanks to Alien Bob's new git repository for tracking the developments in -current as I was able to grab the patch needed for the SlackBuild for cups 2.2.7.

3rensho · 06-08-2018, 01:42 AM

Just for grins I just built cups-2.3b5, installed it and printed some documents. Problem does not appear. I'm running Slack64-current with AlienBob's Plasma5

Edit: Ooops, spoke too soon. Closed Plasma and reopened it and CPU flat out again.