VAXstation 3100/M76 Crashing when Warm
When I first got my VAXen the 3100/M76 was crashing on repeat. At that time it was incredibly hot in the UK. I did wonder if it was the cause at the time, but more recently have had OpenBSD running quite reliably. The past few days I have been experimenting with running NetBSD instead of OpenBSD, since it includes an Xserver (albeit unaccelerated) and isn't statically linked. It led to some posts on
port-vax and here are some of the more interesting messages. I hope they are useful to someone and that Antonio doesn't mind me sharing.
From: Antonio Carlini To: port-vax@NetBSD.org Subject: Re: VAXstation 3100/m76 stray interrupt Flags: replied, seen, list Date: Tue 16 Aug 2016 22:00:50 BST Maildir: /personal/Lists List: port-vax.NetBSD.org On 16/08/16 21:14, Johnny Billquist wrote: > > > I would suggest running VMS on the box for a while, and try to stress > the memory subsystem a lot. VMS have excellent error logging and > management, so if you have any problems, you should be able to get > more details that way. There is also built in tests in the machine, > that could be used. But you'll need to find the manual to work out > which tests to run. Maybe bitsavers have them, I haven't checked. > I had a VS3100/76 many years ago (still have it in fact!) that would exhibit [correctable?] memory errors when things got warm (i.e. in the summer). This was when I was in DEC so with a bit of hunting I found the description of the problem. IIRC there was an issue with the 74LS243 bus transceivers. If you had a mainboard that had a particular batch of a particular manufacturer's variant you were likely to be susceptible to the issue. I'm reasonably sure that I only saw correctable errors and I only saw those because I looked in the error log. But it was a long while ago so there may have been the odd crash too. Anyway, it's entirely possible that NetBSD doesn't know what to do with the interrupt reporting a memory error. So perhaps you ran NetBSD on a warmer day? Or perhaps it is something else entirely? Antonio
To which I responded:
From: Aaron Jackson To: port-vax@NetBSD.org Subject: Re: VAXstation 3100/m76 stray interrupt Flags: seen, list Date: Wed 17 Aug 2016 01:48:39 BST Maildir: /personal/Lists List: port-vax.NetBSD.org > I had a VS3100/76 many years ago (still have it in fact!) that would > exhibit [correctable?] memory errors when things got warm (i.e. in the > summer). I was wondering if this was the problem. When I first got my VAXstation, it was very hot and was hardly working at all, I think OpenBSD was crashing too. It's not as hot now, OpenBSD is stable but NetBSD is refusing to boot. I'll investigate more tomorrow. Very interesting stuff. Thanks! Aaron.
I haven't had a day cold enough to confirm this yet, but will be sure to update this if one comes along.