Dan Price wrote:
> On Wed 01 Mar 2006 at 06:19PM, Rich Brown wrote:
>
>>It prints the header after every 24(-ish) lines of output. I'm looking
>>forward to hearing how this gets used so the output can be polished up.
>
>
> I think the algorithm should be:
>
> If only printing one line at a time (fsstat ufs 1),
> print a header every 10 lines or so.
>
> If printing multiple lines at a time, always print a
> header.
Good suggestion. I'll run this past the CLI folks to see what the
expectation is.
>>BTW, For easier reading (and date/time correlation) try:
>>
>> fsstat -T d {interval}
>
>
> You know that -T is not in the usage message, right?
Well, that's a bug... ;-) (I just submitted it.)
> I think having set/get attr ops available via the -a option is great; I
> just think that as an admin I want to spot the big number, and then
> drill down from there. If attrs were high, I'd use your
> attributes-in-depth option to learn more.
I'll discuss this with the NFS/filesystem team since they were
the driving force.
> I also won some space by merging the name operations; I think that
> new, remove, and name change could all potentially be merged into
> 'name ops'.
These are distinct sets of operations and are useful to keep separate.
I'll bring it up with the other FS folks, though.
> Again, I think the default output is a bit puzzling. There is
> the additional question: what does the default output look like
> in a zone? If the kstats aren't virtualized per-zone, it starts
> to look less and less useful.
The default output looks the same inside or outside the zone. As I
mentioned, I'm not sure (yet) how to separate out the zones part.
Thanks,
Rich
