FW: fatal error: unexpected SIGSEGV in emulator mode

[Fairlight]

On Thu, Dec 02, 2021 at 07:31:41AM -0500, Jose Lerebours via Filepro-list thus spoke:
> Weird, why would it be impossible to migrate on the basis of indexes?

I can't remember the details fully.

I do remember that you cannot concurrently run 5.0 and 5.7 on the same
files.  There's something about it which makes one or the other lock up.
Once you're using one, you're committed to it until a reboot, as I recall.
It's been at least half a decade, so I can't remember the granular details.

> Could they not simply write a script that reads how indexes are build and
> create a script listing said construct and run it to "rebuilt/create" the
> indexes?

It's their wheelhouse, not particularly mine anymore.  If the person who
loves the software and uses it 24/7 says it's not a straight-forward
proposition, why would I question their judgement?  They had every
motivation to upgrade, and it kept getting sidelined.

> Yeah, much pain there, specially if you intended to copy runtime around
> while buying a single license.  :-)   Easy now Mark, I am not implying
> you would do such a thing; we all know better than that.

Honestly, it's not even worth pirating, given the modern offerings out
there you can get for free.

> I do believe that while a fair thing to do, they did not do a good job with
> the LM

Oh, yes...completely fair that your bought-and-paid-for license can just
take a header in the middle of the night because the LM fucks up.
Peachy.

> > And some people are just, "If it ain't broke, don't fix it."  Those are the
> > people for whom SCO was meant.
> 
> Well, now it is broken - lets fix it!

It's exhibiting symptoms.  We don't actually have enough data to know
it's irreparably broken.  It's possibly a question for whatever company
SCO became, or for fP-Tech.  We know the latter aren't going to do squat
to help, so the other company is the better bet, especially since it
seemed tied up with emulation.

Thing is, if it was a static system, and it just suddenly started
occurring, that points to a sudden issue which is probably transient,
if the whole assemblage has been working for ages.  Such things are
often recoverable.  The problem is the paucity of knowledge in areas
that esoteric and arcane.  Most of the people who could have helped
with something at that level in this community are now dead or retired,
unfortunately.  I've little doubt JPR would have been able to help, for
instance.

> There is such a thing as "life expectancy" and it seems that this
> server/components have reached theirs (possibly).

I honestly don't think the OP's data is conclusive enough to say that
with any degree of certainty.

> I am not opposed to being conservative with your $$$ and if things are
> working fine, do not rock the boat but, one must consider the fact that
> you cannot cross your finger and simply hope that any server, or part
> there in, will last you for ever; one must have a plan, a solution ...
> normally, this is address by periodic migration/upgrade to avoid a FATAL
> CRASH.

This did not read like a hardware issue to me.  Not unless the RAM was
dying.  My first thought was memory corruption, based on the error
message.  SEGVs are almost always memory related.  The most common cause
is a bad pointer, or code which results in an invalid pointer when
presented with certain data.  If it's anything to do with hardware, the
RAM should be the first suspect.  That said, if the rest of the system
is working, it seems isolated.  I'd try it again after a reboot, when
all the memory addresses have reset and shifted a bit, and see if it
persists.  If it does, then I'd look at checking the filesystem, as that
could affect data and cause this sort of issue.  If that was clean, or
the problem persists past the point it is repaired, then I'd look at the
RAM.

I honestly think you're being pretty militant about wanting to toss out
the baby with the bathwater.  It sounds like an isolated problem which
may be fixable without upgrading everything under the sun.

> Not to disrespect members of this list but, I venture say that 90+ % here
> would not upgrade solely because of the $$$ - Just look at how often they
> are struggling with "freeware" in lieu of buying Anzio (not affiliated),
> a well known and proven product.

Honestly, I -tried- Anzio, and their font scaling -sucks-.  Then Bob
decided he didn't want to actually fix it because I only owned Lite, and
apparently a customer running 'just' the Lite version wasn't worth
getting his code right.  Okay, fine.  I reverted to Putty and have never
looked back.  The only reason I ever even considered Anzio in the first
place was the background image facility, and it was not worth the other
hell that came with it.  I absolutely hate Anzio now.

> @Cesar Baquerizo.-  I have a CRM you might want to look at and leave your
> troubles behind! ;-)

Jose, you're being a dick, an idiot, or both.  A migration of a whole CRM
on that scale is not something you do as a knee-jerk reaction, certainly
not under the gun.  It would take quite some time to execute, and you don't
start that process from the middle of a crisis...not if you're sane and
have other options which aren't as life-altering.  This has to be some of
the -least- considered advice I've ever seen you offer, and that's saying
quite a bit at this late date.

m->
-- 
Audi omnia, crede nihil.