openssh unix dev - May 2001 - running sshd under AIX 4.3.3 ?

Hi,

If anyone has managed to get sshd to run as a subsystem in the System
Resource Controller under AIX 4.3.3 (a la mkssys), then please let me know
how you did it...

I can mkssys and startsrc it, but it dies immediately, leaving a child sshd
running with another PID than startsrc reported, and lssrc reports sshd
inoperative.
Is sshd a process that should stay in foreground, not forking? Or does it
fork every time to service a connection (excuse my ignorance)?

Douglas Manton, I read your post of 2001-03-16 on this (Re: SIGHUP/av[0]
restart failure (AIX)). Any clues? What am I forgetting?

Thanx,

At 01:36 PM 5/9/2001, you wrote:
 >Hi,
 >
 >If anyone has managed to get sshd to run as a subsystem in the System
 >Resource Controller under AIX 4.3.3 (a la mkssys), then please let me know
 >how you did it...

This works for me, at least on AIX 4.3.2, I don't imagine it would be much
different for 4.3.3.

mkssys -s sshd -p /usr/local/sbin/sshd -a '-D ' -u 0 -S -n 15 -f 9 -R -G
local

Note the -D switch I pass to sshd.




Jason Fortezzo
fortezzo at directlink.net
PGP Key: http://mysite.directlink.net/fortezzo/pgpkey
---
If you have any trouble sounding condescending,
find a Unix user to show you how it's done.
                     --Scott Adams

> mkssys -s sshd -p /usr/local/sbin/sshd -a '-D ' -u 0 -S -n 15 -f 9
-R -G
local
> Note the -D switch I pass to sshd.
Yup.  Previously I had patched the sshd source to prevent it from forking
(a prerequisite of the AIX SRC).

I also wanted the functionality to log if sshd had fallen over.  Using the
"-R" switch on mkssys restarts it, but unless you monitor the PID then
you
have no record.

I have managed this using the "mknotify" method.  You simply run the
above
mkssys without the "-R" and then:

    mknotify -n sshd -m /usr/bin/sysfailed

/usr/bin/sysfailed is a script along the lines of:

    #!/bin/sh
    logger -pdaemon.err -t srcmstr "$1 has fallen over and is being
restarted"
    startsrc -s $1
    CODE=$?
    if [ "$CODE" != "0" ]
    then
        logger -pdaemon.crit -t srcmstr "$1 could not be restarted!"
        stopsrc -s $1
    else
        logger -pdaemon.err -t srcmstr "$1 restarted"
    fi

This logs the failure and restarts sshd.

Now to convince the AIX developers to send a "kill -HUP" to subsystems
controlled through signals using the refresh command!

Best wishes,
--------------------------------------------------------
  Doug Manton, AT&T EMEA Commercial Security Solutions

                E:  demanton at att.com
--------------------------------------------------------
"If privacy is outlawed, only outlaws will have privacy"