RAID Dead Please help (Dell Adaptec Perc 5/i)

[teejay]

What we've done, which worked out the cheapest and easiest solution was to buy a server maxed out with a load of big SATA drives to give us a huge storage area to back up everything onto. We run a full backup of all servers over the weekend and then differential backups on weekdays and take a copy onto removable hdd's once a week. Think it worked out at about £2k in total 3 years ago including a dozen external hdd's so would probably be a lot less now. Much easier and seems much more reliable than tape.

[j17sparky]

Thats my ultimate plan too (great minds think alike eh?)

We actually have an external backup solution on its way for the file server and 1 DC. No plans as yet for backing up the app server which is currently backed up onto an external HD, again something i am not comfortable with.

[teejay]

Yes great minds do think alike! It has been so much less hassle since we moved to this solution as before we were using 5 different tape backup drives attached to different servers. It was a right PITA and extremely unrealiable! What made the switch was a similar situation to what you had, especially when I went through the server backups for the problem server and found that the tapes which appeared to have backed up correctly would not restore. Managed to get the server back up and running with no data loss, but after that scare I demanded a decent backup solution (showing the head a quote from a data recovery agency to recover the server helped!) and also have a much more stringent disaster recovery plan.
Another thing I do is to use VMWare converter to make a VM image of all the key servers during any holiday, so that in a disaster I can quickly get some VM's up and running and restore the latest backup.

[j17sparky]

Anyone know if a HP SDLT320 should be able to restore data at the same rate as it backs it up?

The rated sustained backup speed is 57gb/hour native (uncompressed). So far my restore job has done only 28gb and has taken 3 and a half hours. At the rated speed it should have done knocking on 300GB after decompression.

Looks like im right not to have faith in our backup situation.

[matt40k]

Depends on your load, how the tape is connected to the server.

Normally "Dell" don't give you a dedicated SCSI card for the tape drive. Dispite the huge performance issue.

[j17sparky]

Depends on your load, how the tape is connected to the server.

Normally "Dell" don't give you a dedicated SCSI card for the tape drive. Dispite the huge performance issue.

Opps sorry. Its on its own dedicated SCSI 320MB/s PCI-X card restoring onto a sata mirrored raid. Its a HW raid afaik (never actually looked much at the server tbh), ill look tomorrow.

Surerly though any HD i/o can handle more than 8gb/hour (the speed at which it is currently restoring). The tapedrives SCSI card is certainly not the bottle neck.

[teejay]

If your RAID array is still rebuilding it will slow it down quite a bit. Also writing to RAID 5 is quite a bit slower than reading (but not as slow as the figures you are getting).

[MikeT]

Waking a dead thread I had a similar issue happen this morning with our MD1000. I got a notice from Symantec Backup that the volume was offline and come in this beautiful Monday morning to find 13 of the 15 1TB drives flashing failed on the chassis along with the warning beep. I booted into the PERC5/E BIOS/config and it showed the 2 remaining non-failed drives as OK and foreign configuration. I have never had this issue before but glad I found this thread. I imported the foreign configuration and forced back online the 13 drives one at a time (no particular order). It appears to be healthy but I'm running a consistency check right now. Will probably grab the newest firmware and update. It's been stable for 4-5 years without issue before this on the same server/setup, etc. --Mike

Came in to find the main file server had died. Tried to boot but the logical drive is broken causing the system to see no boot devices.
I go into the Perc Raid setup and find that 2 physical disks of the 4 which form the RAID5 array are showing as missing/forgien copnfiguration.

1 of the disks looks like it is well gone as the warning lights are flashing but the other may be ok. Is there any way of joining these 2 disks (or just 1) back to the array in order to recover teh data?

To make things worse this server may not have a working backup


The RAID is just a plain old RAID 5 with no hot spare.