Okay, so I’m tooling around with the RAID last night and I’m cleaning up the file system after transferring some files over samba. Lets say that I copy over a file called Happy.avi. I get another file called ._Happy.avi. I have no idea why this file is there, and assume that it’s a samba leftover. So I start deleting them with a rm _.* command. This goes okay for a little bit with no weirdness, then all of a sudden, wham input/output error. I can’t ls anything on the drive anymore, and it just shows as empty despite being able to actually change directories I know are there. The only solution I can find is to nuke the RAID. no biggie, drop into SuSE’s disc partitioner and format the RAID. All is well. So I cat the /proc/mdstat just to make sure and notice that one of the drives is missing from the RAID. Umm weird. So SuSE is not adding one of the devices to RAID when it builds it for some reason. It’s there in the raidtab file, and the raid shows up as the proper size, but one of the drives is definitely not there according to mdstat. So I do a raidhotadd and let it rebuild. We’ll see how it is when I get home.
The really weird thing about this is when I originally set up the raid in SuSe, I noticed the same issue. One of the drives was missing, but the RAID was working fine with no issues, so I ignored it. I’m wondering if there is something weird with SuSe’s RAID implementation or if I’m just doing something wrong.
Anyone with unix file system skillz who can tell my why killing those ._ files would make the drive spit back Input/output errors?