Ok, so hereís the dealÖI have an Ubuntu fileserver:
Ubuntu 10.04.1 LTS \n \l
Linux USC 2.6.32-24-generic-pae #43-Ubuntu SMP Thu Sep 16 15:30:27 UTC 2010 i686 GNU/Linux
mdadm - v18.104.22.168 - 15th October 2008
Attached to this server is a 8x1TB MD RAID 5 array housed in a Rosewill enclosure and connected via dual-plane eSATA connections. This array has been sort of a mystery for a while now, and I had a feeling that one of the drives was going bad, b/c for quite some time, when Iíd attempt to write to the array via Samba, even something as small as creating a new directory, the array would freeze up anywhere from 10 seconds to two minutes. Of course, it got longer over the months. What I mean by this is, if I were to create a new directory via a Samba connection, the light (on the front of the enclosure) for one of the HDDs would light up, and the directory wouldnít be created, until the array unfroze, at which point, the light would then go out. In other words, it was pegging one of the HDDs. However, this problem would only occur, when the array hadnít been written to in some time. If I had recently written to the array (say, in the last 10 minutes), this problem would not surface. Nonetheless, I never noticed any out of the ordinary error messages. Since I wasnít sure which drive it was, I didnít want to replace it and mess w/ the array in any way.
Two days ago, I noticed some issues accessing the drive via Samba. I SSHd to the box, and I checked /proc/mdstat. It showed 4 of the 8 drives as missing from the array. The first thing I did was check the lights on the front of the enclosure. In the past, the eSATA cables have come loose (4 of the 8 drive lights on the enclosure will go out), and it will cause 4 of the 8 HDDs to disengage from the array. However, plugging it back in will resolve the issue, although sometimes it has to re-sync. In this case, however, all 8 drive lights were lit up, so there was something else going on.
Next, I checked fdisk, and all 8 HDDs showed up. I also checked dmesg, and there were clearly errors relating to the array, but I didnít see anything specifically related to a drive failure. Regardless, I donít understand why a single drive failing would knock four of them out of the array.
Iíve done a ton of reading, and Iíve found some interesting items. I tried reassembling the array, but the best I could do was add the four missing drives as spares. Iím not sure why that happened, but it was as close as I got. Iíve noticed that the UUIDs are the same on all 8 drives. Iíve also read that itís best to reassemble the array leaving one drive out to prevent a re-sync from occurring. In the Ubuntu forum archive I read a thread that said that itís important to know the exact order of the drives in the array, since the drive letters change. I will say, this is something that happens regularly. My drive letters change all the time, so I was worried about using the --build switch to rebuild the array, b/c I donít know how to find out the original order of the drives in the array.
One other item of note, in dmesg, I searched for info on all 8 HDDs (/dev/sdb - /dev/sdi). All the info was the same for all the drives w/ the exception of /dev/sdi, the drive that I have a feeling is the one thatís gone, or is going, bad. This was the info that was the same for all the drives including /dev/sdi:
[ 9.121457] sd 1:2:0:0: [sdh] 1953525168 512-byte logical blocks: (1.00 TB/931 GiB)
[ 9.121490] sd 1:2:0:0: [sdh] Write Protect is off
[ 9.121492] sd 1:2:0:0: [sdh] Mode Sense: 00 3a 00 00
[ 9.121509] sd 1:2:0:0: [sdh] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 9.121655] sdh:
[ 9.589969] sdh1
[ 9.590127] sd 1:2:0:0: [sdh] Attached SCSI disk
[ 18.415636] md: bind<sdh1>
However, for /dev/sdi, this info was added after the last line:
[ 10.052109] md: unbind<sdi1>
[ 10.685573] md: export_rdev(sdi1)
[ 18.215051] md: export_rdev(sdi1)
[ 18.216447] md: export_rdev(sdi1)
At this point, Iím wondering what my next step should be. Any thoughts would be greatly appreciated. Thank you!