On Thu, Feb 19, 2004 at 05:47:55PM -0400, Daniel MacKay wrote:
> Russel:
>
> > I think that's
> >the primary place where you're going to see differences,
>
> All in hostmon-client.linux, I'm sure.
*nod*
> [...]
>
> One main problem is that it is not built to handle the multiple
> interfaces in netstat. I really don't like the way netstat is done
> anyway - reporting ethernet errors as a percent of transmitted
> packets is a mistake... really it should be something like, "Errors
> since I last looked" and alarm when it's above zero.
Hmmm... that output appears to have counters (rather than percentages),
but maybe my "glance" at it wasn't enough...
In any case, it's not /too/ far a cry from (solaris netstat -in):
Name Mtu Net/Dest Address Ipkts Ierrs Opkts Oerrs Collis Queue
lo0 8232 127.0.0.0 127.0.0.1 446973 0 446973 0 0 0
le0 1500 a.b.c.d a.b.c.d 1246711 0 1465439 1 286 0
...though I seem to remember the linux kernel never increments the
"drop" counter, so it's... well... guess we can look at it anyway and
see if it ever changes.
> In the df department, you need to cover all of the different kinds of
> filesystems you need to say:
>
> df -k -t ext2 -t ext3 -t xfs -t reiserfs -t jfs
Seems to me it would make more sense to just get all "local"
filesystems. Something like a "df -kl" or some such... or is there a
flaw in the logic, like read-only or removeable media or proc/swap
filesystems?
> vmstat looks like:
...guess I'll have to find someone here with a redsplat system and see
how far borked this is, etc. (though I think Tim was offering-up a
patch he did for his old Linux systems, with the caveat that netstat -c
is borked on the platform, anyway -- like vmstat everywhere else)
Russell
--
Russell M. Van Tassell
russell at loosenut com
"Good news. Ten weeks from Friday will be a pretty good day."
|