Strange top(1) output

Dominic Marks dom at goodforbusiness.co.uk
Thu May 12 05:48:44 PDT 2005


On Thursday 12 May 2005 11:39, you wrote:
> On 2005-05-11 13:50, Gavin Atkinson <gavin.atkinson at ury.york.ac.uk> wrote:
> >> Ah, yes!  Good thought.  This could affect the width of the USERNAME
> >> column and push everything too far to the right.  If this is the case,
> >> I'd probably vote for optionally limiting the length of the username
> >> column to, say, 8 columns at most.
> >
> > I would also vote for limiting it to 8 characters.  Even with longer
> > usernames, I suspect 8 characters will be enough to identify particular
> > users (and if it's not there is always they UID view).
>
> That's an option too.  I'm currently trying to get top to display
> something like this (80 columns are used for text, so use a slightly
> wider terminal to view this properly:
>
> ---------------------------------------------------------------------------
>----+ last pid: 11090;  load averages:  1.27,  1.26,  0.86    up 0+01:11:11 
> 03:07:43| 71 processes:  3 running, 68 sleeping					       |
> CPU states: 11.2% user,  0.0% nice, 77.1% system,  0.8% interrupt, 10.9%
> idle  | Mem: 50M Active, 348M Inact, 70M Wired, 20M Cache, 60M Buf, 6340K
> Free	       | Swap: 5000M Total, 5000M Free						       |
>
>   PID USERNAME  PRI NICE   SIZE    RES STATE    TIME   WCPU COMMAND/NTHR   
>    | 4738 root      108    0  1360K   836K RUN      1:28 22.80% find/1     
>        | 638 giorgos    -8    0 13496K  4672K pcmwr    1:33  1.03% mpg123/1
>           | 11062 giorgos    96    0  2428K  1520K RUN      0:00  1.54%
> top/1              |
> ---------------------------------------------------------------------------
>----+

If you don't mind I will share my thoughts on these changes.

>
> This includes at least the following changes (some not visible):
>
>   + The entire header line is limited to the window width too.
>   + The USERNAME column is hard-limited to 8 characters.

This makes me a little uneasy. Its a typical idiom, at least at my Business, 
to have usernames which are of the form 'firstnamelastname', for this reason 
they can be quite long and often the first 5-8 characters will be frequently 
repeated, for example the following contrived names:

rogermoore -> rogermoo
rogermoody -> rogermoo
charlottelane -> charlott
charlottedaniels -> charlott

If there are many processes (samba, imap, etc) running as these users (also 
typical in my environment, and I guess elsewhere too). If I use top to view 
the active processes now I will be unable to tell what is really going on. I 
know that I could run top and just display the UIDs, or I could use ps, but 
top is great for seeing the status at a glance, so loosing this information 
entirely in top would make life harder. Also, imagine a system where you have 
200 users with UIDs ranging from 2000 to 2200. Picking out and distinguishing 
between lots of these numbers from a moving top display, is likely to be a 
very error prone task, no?

I suppose it could be argued that using usernames of that format is a poor 
choice on my part, and if that is the consensus of opinion then I'll have to 
look at fixing my own setup. I use long format usernames because it do not 
wish to have to remember that rm5, rmoore (many other possible ways) is a 
particular person.

If this behaviour could be turned on and off, I'd be very happy.

>   + The THR column is displayed as /1 after the COMMAND, like the
>     prstat(1M) command of recent Solaris versions.

I like this.

>   + The CPU/WCPU columns occupy the same space and can be toggled with
>     the 'C' keyboard command.

I like this too.

>   + When UID numbers are displayed, hitting 'u' will read a UID instead
>     of a username.
>   + When the view is toggled between processes/threads, the NTHR part
>     becomes the thread ID of the particular thread.

Okay, not really sure what this will look like to me but no need to explain 
I'll wait until they hit -CURRENT and see for myself.

> Hopefully, I'll have these changes running on CURRENT before the weekend.
>
> If no strong objections are voiced for any of these changes, I'll test
> it on CURRENT for a while, then ask for approval of a commit to HEAD and
> merge it to 5-STABLE after it's been tested enough on CURRENT.

Thanks for your work !

> - Giorgos

-- 
Dominic
GoodforBusiness.co.uk
I.T. Services for SMEs in the UK.


More information about the freebsd-stable mailing list