SCO Open Server 3 (SCO report)

[Tim Rice]

On Fri, 26 Jan 2001 mouring at etoh.eviladmin.org wrote:

> On Fri, 26 Jan 2001, Aran Cox wrote:
> 
> [..]
> > If I don't do that I get:
> > checking for PRNGD/EGD socket... ./configure: test: argument expected
> > 
> 
> This has been brought up once.  It's choking on "if test -S
> $egdsock; then".  
> 
> The test verifies that it's detecting a Socket.  And it seems that SCO is
> missing -S... (Hmm.. I should check NeXT it may also lack a -S)  I
> understand why -S was selected, but I don't know how many older platforms
> support -S without GNUaifying them.

All the UnixWare platform also. Probably most System V based machines
not counting Solaris.

> 
> I hate to change it to -e  ..  Only other option I see is doing a

The -e option of test is not portable either.

> "test ! -f $egdsock" .. At least would at least get us close to
> what we want.  Assuming -f is portable and I believe it is.
> 
> [..]
> > Anyway, on to the USE_PIPE question.  A while back I had a problem
> > with 2.1.1 and later editions where commands run remotely would die like so:
> > 
> > ssh -n -l root tignam3b ls
> > Received disconnect: Command terminated on signal 13.

This worked for me on 2.2.0p1
ssh -n -l mt sco504 ls
mt at sco504's password:
Main.dt
Personal.dt
bin
mods
mt_mods
sco504
trash.dt

... installs Jan 25 CVS .....
tim(trr)@uw213 10% !!
ssh -n -l mt sco504 ls
mt at sco504's password:

 Hmm, Stuck here.
And ps on the SCO 504 machine shows
   RUSER   PID  PPID   %CPU  VSZ        TIME     TTY COMMAND
    root 12184 12138  92.64 1136    00:00:01       ? /usr/local/sbin/sshd

Looks like some more research is needed.

> > 
> > I don't recall the details but I found that defining USE_PIPES cleared
> > up that issue.  Now it seems that the configure section for sco has been
> > broken out into sco3.2v4 and sco3.2v5 and v4 has USE_PIPES defined
> > and v5 doesn't.  I don't have the previous problem with commands 
> > terminating on signal 13, nor do they seem to be hanging, in fact
> > everything appears to be just fine.  Can anyone explain what has
> > changed in this regard and why?
> > 
> 
> I have not paying too much heed to SCO until recently (mainly
> because I don't run it =).   Tim, is there any reason why USE_PIPES is not
> defined for sco3.2v5?  Or should I add it in?

Before discovering opsnssh I whiped up some portability fixes for 
the ssh.com version. I don't think I needed USE_PIPES on Open Server 5
there (Just Open Server 3) and as it turend out, I didn't need USE_PIPES
on Open Server 5 on the openssh-2.2.0p1 version I patched.

It's possible I didn't uncover the problem yet.

... does more "ssh -n -l mt sco504 ls" tests with the CVS version .....
Looks like it's working 8 out of 10 times.

If Aran is having problems and USE_PIPES fixes it, I say put it back in.

> 
> - Ben
> 
> 

-- 
Tim Rice				Multitalents	(707) 887-1469
tim at multitalents.net