Hi,
The original question:
On our 4.0F system, an executable called exclOpen (in /usr/sbin) is used in
the rcmgr script to implement a locking mechanism. I can't find a man page
on exclOpen, and it is not on our 4.0D system. Since it's also not on a
Solaris system that I have access to, I assume that it is a fairly recent
Compaq-specific utility. It's pretty easy to figure-out how it works, but
I was curious for more info as it may be a useful utility. (By the way,
the man page for open(2)/creat(2) mentions an O_EXCL switch, so I guess
that's a key part of exclOpen.) Any comments?
Thanks to the following for their quick replies. Each one was educational
-- that's hard to beat! And it's always great to hear from folks from the
Compaq side of the fence. It's been said many times before, but this is a
GREAT group!
Regards to all,
Peyton Bland
University of Michigan, Dept. of Radiology
------------------
>From Colin.Walters_at_compaq.com...
If it was in V4.0x base OS, that script is gone from the V5.0 and later
base OS, unless:
o You are running a cluster and the script is cluster-specific.
o It was put there by an optional subset or an application.
I cannot turn up any reference to exclOpen running a string search on
the entire docset and manpages. However, the string O_EXCL is found
in the following places:
open(2)
Messages (Guide to Realtime Programming)
Shared Memory (Guide to Realtime Programming)
Semaphores (Guide to Realtime Programming)
mq_open(3)
sem_open(3)
shm_open(3)
proc(4)
mc(7)
ksh93(1CDE)
dbopen(3)
Apologies for the lack of info. It is not unusual for temporary
utilities to be undocumented when we know they are to be short-lived.
Consequently, it is risky to make use of any scripts and utilities
that are not documented. They might vanish in the next release.
I know that rcmgr and rc.config have been greatly modified in the
last few releases (to support clusters) and this might have been a
temporary part of those rc* changes.
----------------
>From Dr. Thomas P. Blinn, tpb_at_zk3.dec.com...
You got the idea. It should never have been put into a "standard" place,
it should have been somewhere like /usr/lbin. Its only purpose is to
get a real lock on a file, since "rcmgr" is a shell script and thus has
no way to get exclusive access (through normal shell mechanisms) to the
files it needs to update. And if it doesn't have exclusive access, it
can screw up the files. (Creating a file and checking for the presence
of that file doesn't work; you can't avoid race conditions in shell
logic, you need something more primitive.)
Think "major hack" and you'll be on the right track. I could probably
get the source for you; I'm sure it's encumbered only by us, not by any
external party, but it's probably not generally useful as written.
----------------
>From John P Speno <speno_at_isc.upenn.edu>...
Looks like it takes an argument as a filename to create and tries to open
that file with O_WRONLY|O_CREAT|O_EXCL flags set.
If that file exists, or it can't open, it gives you an error message.
That's about it.
P.S. I used a system call tracer to see the arguments to open().
Received on Mon Jan 22 2001 - 16:13:09 NZDT