Re: Problems with scheduler on SCO

[Matti Aarnio <mea@nic.funet.fi>]

> I cannot make the scheduler work properly: if a message is scheduled
> for delivery to multiple hosts, the scheduler starts one transport
> process that is completed successfully.  According to the log,
> the scheduler starts more transport processes, but I never see them
> in the `ps' display, and the message is never delivered to the
> remaining addresses.  Also, there is suspicious diagnostic in the
> log:
> 
> ====
> 14 processed: 12038/119         deferred can't reset uid to root
> 14 processed: lockaddr: expected '~' found '+'
> Misformed diagnostic: lockaddr: expected '~' found '+'
> ====

	"can't reset uid to root" ?

	Does your system not have seteuid(), or setreuid() ?
	(When I heard about SCO platform as attempted target,
	 I was immediately suspicuous recalling my own experiences
	 with ISC's SysVr3 platform..)

	The system does require reversible euid setting, without it,
	there are Serious Kludges available in form they were made
	for some XENIX, but opening kmem and poking process uarea
	is not (IMO) very nice thing to do..

> It seems that the scheduler has problems communicating with the
> child processes...  Any advice?

	Those errors are indicative of having the   scheduler.conf
	somehow misconfigured.  Please send ME a copy of your current
	scheduler.conf,  and I will try to figure out, what you missed.

> I am including the message from the `transport' queue for reference.
> 
> Thanks,
> Eugene

	/Matti Aarnio	<mea@nic.funet.fi>