I have installed raidtools-0.90 (19981214 - patches also) I am performing some test. I have some space on an internal hard-disk, and an external zip drive. so I have placed 1^ disk on hda1, 2^ disk on sda1 (the zip), so I can eject it during work. mkraid ok, restart after a powerfail ok (it resyncs transparently, good). then I tried to eject disk. raid runs in degraded mode, ok. but then I am not able to reinsert sda1 on disk array, I remain in degraded mode even after a reboot or a powerdown. in my logs after reboot I have: Jan 20 00:06:43 localhost kernel: hda1's event counter: 00000003 Jan 20 00:06:43 localhost kernel: sda1's event counter: 00000001 Jan 20 00:06:43 localhost kernel: md: superblock update time inconsistency -- using the most recent one Jan 20 00:06:43 localhost kernel: freshest: hda3 Jan 20 00:06:43 localhost kernel: md: kicking non-fresh sda1 from array! Jan 20 00:06:43 localhost kernel: unbind<sda1,2> Jan 20 00:06:43 localhost kernel: export_rdev(sda1) Jan 20 00:06:43 localhost kernel: md0: removing former faulty sda1! why it "kicks" non-fresh sda1 from my disk-array, and it doesn't resync ? the only way for reinstall sda1 on raid is mkraid --really-force /dev/md0 !!! I thing thet this is not the right way. in RAID5.HOWTO I read: --------------------------------------- * How do I recover from a disk crash ? . . When you plug in a new disk (or in my case, re-connect the disconnected disk after writing a lot more information to the RAID), you should see the RAID subsystem automatically beginning to reconstruct the information that was erased. but I don't see anything about the automatic beginning to reconstruct .. neither after I reinsert my sda1, neither at reboot. I tried raid-1 and raid-5, results are same. I tried even with a newly-formatted zip, but nothing of good. where is the problem ? In old raidtools there was mkraid --only-superblock - now there is nothing of similar. what is the correct procedure after substitution of 1 disk ?
