Re: DESCHALL runs wild when run under nohup?

C Matthew Curtin (cmcurtin@research.megasoft.com)
Mon, 5 May 1997 14:35:48 -0400 (EDT)


>>>>> "Lee" == Lee Sheridan <lsherida@mail.bcpl.lib.md.us> writes:

Lee> Because BCPL has a limit of 3

(Maybe-obscure programmer joke coming...)
Gee, I thought people quit using BCPL about 20 years ago. :-)

BOFH> When I came in this morning I found the UNIX host running very
BOFH> slowly. The culprit was your proccess "deschall-sparc
BOFH> keymaster.verser.frii.com", which had been running since 11:40 PM
BOFH> Sunday. When I discovered it, it had used a tremendous amount of
BOFH> processor time and was still growing rapidly. I killed it off at
BOFH> about 8:35 AM.

I tend to doubt this. What hasn't been stated is what the priority of
the process was. Likely, he was looking at the process table, and saw
something like

cmcurtin 7987 11154 68 Apr 18 pts/4 23760:16 ./deschall-sparc keymaster

Of course, he's going to see that it's taking all kinds of processor.
The "growing rapidly" is also a given, depending on how you define
rapdily, since every *cycle* that isn't otherwise used will be used to
chew the keyspace.

Some additional information, like the nice(1) value of the process,
compared to other stuff, or whether it was using a lot of memory, etc.

Of course, I don't have all of the facts, but it smells like you're
dealing with more of a problem with perception than actual system
usage.

One possibility for real "performance problems" might be that the
deschall client is running with a nice value of 4, and that be the
same, or higher, than what a number of processes that they use are
doing. (It's possible to configure Solaris to reduce the priority on
processes the longer they've been running. Hence, deschall might be
at 4, and so might some other stuff they've been using, or they might be
at 5, etc.) You might consider doing a

$ nohup nice -10 ./deschall-sparc foo bar baz

And seeing how the system "feels" comparatively. Alternatively, find
out what other processes feel pokey, and compare nice values. (Might
be useful to do this whil BOFH is around to see how things look...)

Hope that helps.

-- 
Matt Curtin  Chief Scientist Megasoft Online  cmcurtin@research.megasoft.com
http://www.research.megasoft.com/people/cmcurtin/    I speak only for myself
Death to small keys.  Crack DES NOW!   http://www.frii.com/~rcv/deschall.htm