On 25Nov2020 22:33, tech-lists <tech-li...@zyxst.net> wrote:
>On Thu, Nov 26, 2020 at 09:01:03AM +1100, Cameron Simpson wrote:
>>A way to check this would be to have another window open running:
>>
>> strace -p pid-of-idle-mutt-process
>>
>>Get that ready. Wait for idleness. Resume. See where it stalls.
>>
>>If that is hard to observe interactively, strace has options for
>>including timestamps in the trace output, you could do the resume and
>>then scroll back looking for the stall in the timestamps.
>
>Unfortunately strace (do you mean devel/strace)
Probably. I don't have a FreeBSD box to hand. Is yours a Raspberry?
>appears to be broken:
>
>root@rpi4:/usr/ports/devel/strace# make
>===> strace-4.5.18_1 is marked as broken: Uses procfs-based process debugging.
>*** Error code 1
>
>Is there an equivalent tool?
Have you got ktrace and kdump? Less interactive, but IIRC the process is
to ktrace your mutt command and kdump the resulting log file.
See this:
https://www.unix.com/man-page/FreeBSD/1/ktrace/
Note that you can trace particular stuff with the -t option. Options "i"
and "n" and "c" look promising for what I'm imagining is your problem
(hard drive spun down, slow to spin up).
You can attach ktrace to a running process with -p, so "ktrace -p
pid-of-mutt".
So you could:
- start mutt, let it idle
- ktrace -p pid-of-mutt
- resume mutt, wait for it to wake up
- stop the ktrace
- use kdump on the ktrace,out file, see where mutt spent its time
Cheers,
Cameron Simpson <c...@cskk.id.au>
