Hi Chris, hi zhenwei,
Chris Hofstaedtler wrote:
> > The irqtop(from util-linux) command has better user experience:
> > - aggregate interrupt information seems clear on a morden server(128
> > CPUs or more).
> > - sort by several rules, include IRQ, TOTAL, DELTA and NAME.
> > - specify cpus in list format to monitor.
> > - specify output columns to print.
>
> thanks for the reminder. For irqtop, I had some discussion with the
> maintainer of the current irqtop package (CC'ed now). I cannot
> remember if we came to a conclusion though. Axel, maybe you can
> remind me...
Since you're asking, I allow myself to cite my reply to your inquiry
with me back then (June 2021):
---------------------------------------------------------------------
Hmmm, do they do the same? Can I test that irqtop from util-linux
somewhere?
Since people seem to expect the irqtop tool from util-linux, I see
multiple options:
1) If the irqtop from util-linux is clearly superior: Drop building
the irqtop package from src:iptables-netflow and let util-linux
generate a transitional package. (Versions should be no problem
with 2.6 < 2.36.)
I more or less built that binary package, because that tool was in
the upstream sources and no such feature was present in Debian so
far and I didn't want to pull in ruby just for a DKMS kernel module.
2) If none of them is clearly superior and they have different feature
sets, using the alternatives system might be an option.
Since I expect both to be just TUI tools without having an API
being used by other tools, different commandline options should not
be an issue here.
3) Rename one of them, e.g. to irqtop-nf or irqtop-ul or so. (Renaming
both of them will be needed for variant 2 anyways.)
In case you intend to add lsirq for bullseye, you could also add
irqtop as irqtop-ul or so (i.e. variant 3b), too. That shouldn't cause
any disturbance IMHO.
---------------------------------------------------------------------
As far as I can see, I didn't get a reply back from you on these
suggestions of mine. Maybe my mail fell through the cracks. But I
think we should take the discussion up again, probably in this bug
report.
Another point which comes to my mind now is that it might make sense
to rename the current irqtop package to irqtop-nf (or irqtop-ruby)
just to make clear that it does not contain the irqtop tool from
util-linux.
Regards, Axel
--
,''`. | Axel Beckert <a...@debian.org>, https://people.debian.org/~abe/
: :' : | Debian Developer, ftp.ch.debian.org Admin
`. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5
`- | 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
