🚀 go-pugleaf

In article <1993Jan13.175210.374@serveme.chi.il.us> greg@serveme.chi.il.us (Gregory Gulik) writes:
>In article <C0sJxE.6K2@ddsw1.mcs.com> karl@ddsw1.mcs.com (Karl Denninger) writes:
>>
>>You don't push serveme hard enough on the network.
>
>That's very ture.
>
>>There's something about IP and Suns.  For all the ballyho about "the network
>>is the computer" these things get mightily upset if you actually try to USE
>>them that way.
>
>Hmmm..  We seem to have over 1000 Suns at work, and I've never
>seen this "scrambling" problem.

You also have "real" file servers there (Auspex makes nice ones ;-)

Trust me, if you work a Sun hard enough on the network, it will fall down.
Sun has gotten much better as of late, but they're still not that wonderful. 
The trick is to know what "permitted" things don't work, and not to do any 
of them.  If you do, you find out real fast how to do a "STOP-A" :-)

>>I love Sun's "patch of the hour" club; their "jumbo" patches are quite a
>>treat.  If you have ever run a large network of Suns, trying to do things
>>that would seem natural (anyone can sign on from anywhere, automounting
>>resources, etc) you find that the stability of their entire package leaves
>>something to be desired.
>
>Really?  I do that regularly at work.
>
>Actually, our network is a monster NFS network with Suns, SPARCs,
>NeXTs, MIPS's, MPC's, etc...
>
>It all seem to work fairly well.

The automounter which Sun ships is buggy as hell.  I can come up with a
hundred ways to make it fail.  AMD is considerably better; it doesn't hang
the machine or processes on it nearly as often.

You obviously don't use the lock daemon much; if you did you would have a
profound understanding of what I speak.  The road to hell starts with
rpc.lockd, especially if you're nutty enough to have diverse machines on the
network (as you do).... ;-)

>>Yeah.  It seems rather, uh, insane to run NIS unless there is more than
>>one machine sharing UIDs on Ethernet in the clout "family".
>
>No kidding.

Yep.  NIS has its problems too, not the least of which is that it is a
wide-open security hole.

--
Karl Denninger (karl@ddsw1.MCS.COM, <well-connected>!ddsw1!karl)
Data Line: [+1 312 248-0900]