[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Re. Re: high cpu load systemd-journal, nobody logs



	Hi.

On Thu, Jul 02, 2020 at 07:13:11AM +0200, Nico Becker wrote:
> hello,
> i am again. i tcpdump the port.
> > tcpdump -pni any -c1024 -s0 -w /tmp/journal.pcap tcp port 19531
> i check the capture size after the cpu load is at 100%.
> but the size doesnt increase after the systemd-journal use 100 %cpu.

Ok, scratch that one hypothesis. It was a good one though, it just
contradicted the reality :)


> => i think it isnt the connection to the "client".
> i disconnect the ethernet interface and the cpu load is still at 100%
> 
> i doesnot know howto get more debug information from the process,
> why or what happened if the load is such heavy.

There are ways, but most of them require some sort of control on the OS
(UART console will do). Namely, it's "ltrace" or "perf top".
The idea is simple - you see a syscall that troubles you. But what
journald library call does it correspond? Both "ltrace" or "perf top"
should answer that, but both might as well just hang (there were bugs
about both for both armel and armhf in stretch).
Worth a try, nevertheless.

Reco


Reply to: