Theodore Wynnychenko wrote: > Hello: > > I hope these questions are not off topic. I have been working with the 4.4 > release for the last few months (learning something new keeps brain cells > alive). I got an old P2 or P3 (?) up and running as a firewall at home. It > appears to be very stable, however, every 1 to 3 weeks or so, I get an email > from cron telling me that a script had a "segmentation fault" (I set up some > basic scripts, the one that seems to have this issue runs every 5 minutes, > scans authlog for failed ssh attempts, and enters those ip's into a > blacklist table for PF). So, it's not a big deal, because 5 minutes later > it runs again, and is fine. > But, I was wondering, how concerned should I be about this? I understand a > seg fault is pretty vague, and can mean any of a bunch of things (e.g.: a > corrupt memory cell). So, this is not a "critical" system, but then again I > don't want to come home one day and spend time replacing it before I can get > my email. Similarly, it doesn't happen often (once every 1-3 weeks), so, I > don't want to waste a bunch of time ripping the computer apart trying to > find the source of the error (if it's hardware), if the system is otherwise > stable, and this doesn't pose a substantial risk of failure. Should I be > worried about this or not?
Personally, I would make slow, graceful moves towards replacing the machine. I'm guessing you are getting these out of shell scripts or perl scripts..things that shouldn't be giving segmentation faults. (ok, nothing should be, really, but I suspect if ksh or perl had bugs you were exposing, they probably would have been noticed before. Home-grade firewall machines are found on curbsides on trash day, so I really wouldn't spend a lot of time trying to figure out what part it is that is bad... AFTER you replace the system, you can spend some time trying to figure out what is broke...but it is not worth the downtime, I suspect. I've got a few AMD K6-2 systems, all of which can't compile the system when run at their rated speed, all of which did just fine if I knocked down the bus speed a little below rated speed (83.33MHz rather than 100MHz), and even kicked up the multiplier to get much of the "lost" performance back. > > Second question: Given my success with the first system, I started working > on another old computer (I don't know, maybe a fileserver). While > installing and setting up 4.4, I added "powerdown=YES" to rc.shutdown, so it > would automatically powerdown on "halt." However, when I have the system > halt, it shuts down cleanly, gets to the "powerdown" stage, I hear the disks > stop spinning, but, the power supply fan, and the CPU fan keep running. > Also, a video signal is still present (the monitor does not go to sleep). > So, I hit the power switch for a couple of seconds, and it powers off > completely. APM is present and running. This system is an old HP Pavilion > 8580C (an ASUS motherboard, I guess). I thought maybe there was a problem > with the power supply, so I replaced it with a different ATX power supply, > but same problem. I would guess this is a problem with the motherboard's > APM, but I was wondering if there is any sort of workaround for this? try looking at /etc/sysctl.conf, #machdep.apmhalt=1 # 1=powerdown hack, try if halt -p doesn't work My old IBM Thinkpads are very fond of that. No promises for you, though. Experiment with "halt -p" before sticking things in rc.shutdown... However, I've also seen defective boards that just didn't power down the power supply...even in the OSs they were orginally sold with, and some that just don't try. File servers that don't power down on their own? I really suspect you won't care much once it goes into production... Nick.
