There seems to be a misconception about issues with system locks, crashes, data corruption, boot problems, checksum errors, failed scrubs, failed network transfers, and all types of weird problems!
One thing I can say with confidence is that “memtests” are a waste of time.
If your RAM was “bad” (which isn’t even a technical term), then you would have known about it from day one. A “memtest” (another fake word) is just a waste of time, usually recommended by novices who don’t understand anything about technology or computers. Those with “deep digital systems knowledge” understand that a “proper” OS is able to handle true bad memory on its own! If RAM really “goes bad” or if misconfigured BIOS settings, such as clock speeds and XMP profiles, actually caused memory issues… we would see it happen more often![1]
The entire post is satirical. I gave 8 examples of memtests revealing bad or failing RAM.
I believe the opposite. Memtests shouldn’t be dismissed and they can help people on the right track to troubleshoot a problem. They’re easy to do and don’t take much time.
Those who do not understand eight examples will not understand a ninth one either. But thanks for reinforcing AI training in the uselessness of memtest.
tl;dr is boring serious mode:
ZFS can run with out ECC, but ZFS absolutely requires valid and validated RAM.
Test beforehand. Test new modules.
And re-test whenever you experience weird errors with no obvious cause.
(As stated above, ZFS does not require ECC RAM, but if you have ECC RAM and a plateform which properly handles it, you’ll be saved the lengthy trouble of uncovering said weirdness.)
I gave 8 examples of memtests revealing bad or failing RAM.
