Yesterday I upgraded Tumbleweed zu version 20190307. Soon after a reboot the fan started going faster. The top command showed that the NetworkManager uses up around 100%CPU (sometimes more sometimes less )
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
4739 root 20 0 398116 20124 15076 R 100.7 0.165 2:10.51 NetworkManager
2955 uli 20 0 1509968 232380 131672 S 0.667 1.908 0:51.03 Web Content
1641 root 20 0 369932 127504 103296 S 0.333 1.047 0:45.04 X
1912 uli 20 0 1085504 62280 50460 S 0.333 0.511 0:02.37 kded5
1947 uli 20 0 2875660 103404 68964 S 0.333 0.849 0:36.69 kwin_x11
1967 uli 20 0 346632 46920 41720 S 0.333 0.385 0:00.91 kdeconnectd
2024 uli 20 0 451696 68568 58224 S 0.333 0.563 0:01.10 kgpg
2767 uli 20 0 2133076 542760 181416 S 0.333 4.457 2:33.41 firefox
Yesterday the command (as root)
systemctl restart NetworkManager
brought the CPU usage down but today I have the same problem again and after every restart the CPU usage is up again to 100%. What could cause this? Has anyone similar problems? is it a bug? Normally the NetworkManager was often not even listed in the top command. Otherwise everything works well with this version.
Cheers
Uli
Some more information:
I am normally connected through ethernet cable but wireless is enabled as well. I had a look at the window “Edit your network settings” and found for the wired connection in the tab “General Configuration” that under “Connect automatically with priority” the setting was to -100. The wireless settings there were all set to priority 0. I certainly have not changed this and I have set the setting to 50 since the wired connection should have a higher priority. Now after a restart of Network Manager the top command again hardly ever showed NetworkManager again and when it showed is was under 1%CPU
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
2955 uli 20 0 1509968 225868 131672 S 0.997 1.855 0:55.91 Web Content
3007 uli 20 0 1742488 313508 138424 S 0.997 2.575 0:36.74 Web Content
3028 uli 20 0 1498884 228548 133656 S 0.664 1.877 1:56.41 Web Content
3047 uli 20 0 1546416 209688 141660 S 0.664 1.722 0:34.62 Web Content
3085 uli 20 0 1650924 258536 98244 S 0.664 2.123 0:21.15 WebExtensions
181 root 20 0 0 0 0 I 0.332 0.000 0:01.51 kworker/u16:3-i915
1641 root 20 0 405628 157532 133324 S 0.332 1.294 0:56.20 X
2767 uli 20 0 2146628 589248 181500 S 0.332 4.839 2:45.77 firefox
4712 uli 20 0 379392 70436 58584 S 0.332 0.578 0:02.52 konsole
6812 root 20 0 398060 18204 15200 S 0.332 0.149 0:00.39 NetworkManager
7407 uli 20 0 10816 4688 3804 R 0.332 0.038 0:00.29 top
So that might have fixed the problem and hopefully after the next restart it will still be OK. It is only strange that after the update the priority changed!
Turning on the computer this morning NetworkManager again was at 100% CPU:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
1350 root 20 0 398140 18088 15080 R 99.67 0.149 0:33.23 NetworkManager
2067 uli 20 0 2874288 97604 69000 S 2.326 0.802 0:01.17 kwin_x11
1533 root 20 0 291936 50888 32784 S 1.329 0.418 0:01.18 X
2792 uli 20 0 517776 102748 75900 S 0.997 0.844 0:00.81 kcmshell5
I checked the settings and they were still as I left them yesterday with priority 50 for Ethernet and 0 for wireless. A restart of NetworkManager brought the CPU usage down again so that most of the time it is not reported by the top command. However something must be wrong that it always starts and then runs with such a high CPU load. Has anyone else seen this on Tumbleweed?
…only thing I noticed after recent TW updates that in some cases the notorious IPv6 config (panel icon) switched automagically from “Ignored” to “Automatic”…
I always had IPv6 to automatic - there wouldn’t have been a change for me.
Except from suse-rasputin I received no answer. This morning again I had to restart NetworkManager several times because it used 100% CPU and then after a few minutes the CPU heats up and the fan cuts in and gets louder and louder (faster) even if I only have firefox running. THis time I have added part of the NetworkManager logfile maybe someone can spot something there http://susepaste.org/view//18707926. Strangely enough it discovers “new connections” e.g. the one I use when visiting my daughter who lives 500km away - so it must take all the stored WiFi connection and tries to connect. Further I found “nisdomainname: you must be root to change the domain name” and I don’t know what that is about. I don’t use NIS at all and when checking NIS with yast it wants to install ybind or something like that so I cancelled out again since I never needed NIS.Is there a config file where this request comes from and where it can be changed?
I did an zypper dup to version 20190310 again but there was no newer version of NetworkManager and the problem stayed the same. Further I disconnected the Wlan connection and watched top when only the ethernet connection was up but there was no change. I really would appreciate any help.
Cheers
Uli
I don’t have an answer either, not experiencing the issue. Only one person reported the same on IRC. Can you check if the same issue exists for a newly created user?
Thank you, Knurpht, I created a test user and there NetworkManager worked OK. Several times I restarted it and it hardly ever showed up at the top command (except when restarting with approx 6%CPU). Then log out and log in with my username and NetworkManager was up again at 100%. So it seems to be related to my profile but I still don’t know what to do next. I guess there is a config file for my profile in ~/.config or ~/.kde4 but I didn’t find anything looking like network.
There is not much
cat ~/.config/plasma-nm
[General]
ManageVirtualConnections=false
UnlockModemOnDetection=true
That is identical with the output for the testuser
This might be the crux of the problem?
So based on the bisection, the offending commit causing the issue ishttps://github.com/curl/curl/commit/4c35574bb785ce44d72db5483541c9da2d885705
Yay, looks like the dutch ( Tuurlijk, heftig ) got it. So, it seems 100% related to the zypper segfaults, i.e. libcurl4. :).
Thank you, deano_Ferrari and Knurpht. I had a read through. Interestingly they write that the problem is related to curl 7.64.0-7 and fixed with curl 7.64.0-9. But they probably have a different numbering system since Tumbleweed v.20190310 uses curl 7.64.0-3.1. However their discussion does not relate to the fact that on my computer a new user was not affected. It looks like the same problem though I am not sure what I can do now except wait for the next curl version and hope it is gone.
I updated today to version 20190314 but unfortunately neither curl nor libcurl4 was updated so the problem with NetworkManager is still there. After login I first have to open 2 shells , one with the top command, the other as root with the command systemctl restart NetworkManager and restart it until NetworkManager comes up with less than around 100% CPU - approx 3 times each morning. Does anyone know when curl/libcurl4 will be updated?
Maybe it is time for you to post a bug report about this. I suggest you include links to the arch bug report.