Skip to main content ↑ ↓
Topic solved
This topic has been marked as solved and requires no further attention.
Topic: zaplocked installation recovery SELF-RESOLVED? (Read 1884 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Re: zaplocked installation recovery SELF-RESOLVED?

Reply #30 –
re "I would like to hope that all Linux systems always get ready for the power-failure about to strike" a COW file system will be ready but others might not be, including the popular ext(n) ones. COW means Copy On Write, data is written to the drive then Copied into it's final place, so if you shut down randomly the drive will always be in a consistent state because you never have the situation where there is some half finished write at shutdown, if the copy failed it won't match the original., if the original failed there won't be a copy to compare. This is slower in theory but isn't in reality, nothing that I have ever noticed using BTRFS, perhaps it has other improvements which compensate. I think it actually just treats the journal checksums like this, so not all data is copied but any incomplete stuff can be identified on reboot. It's usually the case a hard power off on a non-COW fs will be fine too, but not if you hit the wrong microsecond. RAID on multiple drives would be even more reliable but obviously not so practical on most machines without hassle as they have only one drive.