4 ms·
I had similar issue with BTRFS thanks to new faulty DDR4. - CRC checksum for some blocks on fs was wrong - I discovered faulty DDR4, replaced it, without know
by t344344 3y ago
I had similar issue with BTRFS thanks to new faulty DDR4.
- CRC checksum for some blocks on fs was wrong
- I discovered faulty DDR4, replaced it, without knowing FS corruption
- system started randomly kernel panicking
- no msg in logs (fs died with kernel)
- eventually I experienced kernel panic while at computer, and seen msg on screen (broken checksum or something)
- there was no way to repair filesystem. I had to recover as much as data, throw away some files and start over on fresh
- I use BTRFS snapshot streaming in backups, wrong checksums propagated to backups. I had to restore from 1 week old backups
- Since than I use RSYNC backups as well
- And I test all PCs with memtest monthly overnight. It is also a good stress test for CPU.
- nicolaslem 3y ago> And I test all PCs with memtest monthly overnight. It is also a good stress test for CPU. I think that the consensus in the PC enthusiast community is that memtest is quite dated and not very good at finding instability in modern hardware.
- tcrenshaw 3y agowhat's the community recommendation these days?
- Marsymars 3y agoI’m not particularly up-to-date on that subject, so someone else can probably provide more info, but FWIW PassMark’s MemTest86 has had ongoing development, while the open source MemTest86+ was effectively a dead project for nearly a decade before some recent activity - so there’s a fair amount of difference between the two.
- t344344 3y agoI am not sure what consensus is, but I use this: https://memtest.org/ https://memtest.org/