I recently managed to find an inexpensive source of RAM for my trusty old server, nexus.zyp.org (an old IBM IntelliStation, dual Pentium Pro 200, formerly with 192MB of RAM, now with 448MB). I bought a 256MB DIMM for it (it was beastly trying to find a decent price for this stuff. . . it takes ECC EDO 168 ping DIMMs, which were only really used for high end PPro servers/workstations, and a few early PII's, because then SDRAM came out, and due to it's much greater speed, was quickly adopted), and installed it Friday night.
Within an hour, the machine had crashed, with page fault errors. I rebooted it, went out to get something to eat, and returned to find it down again. With page fault errors. I rebooted it again, and this time it didn't even make it through loading the kernel before just hung.
At this point, I was getting understandably upset. I rebooted it one last time, after letting it set for a few minutes, and this time dropped it into single user mode. I ran memtest on all available memory, and it went through three iterations without finding a single problem. That took almost 9 hours.
At that point, I threw it back into multi-user mode, started things back up, and figured I'd see how things went, since it'd already been up for 9 hours without problems.
When I woke up today, it was still going. Right now, it's at 24 hours, 14 minutes.
So, should I be worried over the crashes?
Or happy that I haven't seen any problems in the past 24 hours?