software raid 5 drive failure and rebuild

[slaterson]

i have a large raid 5 array (software, not hardware). it consists of 8 drives. this morning i heard a funny noise coming from the box and decided to check mdadm to see the status of my raid. turns out /dev/hdb1 has been removed from the array.

i have a spare ready to replace the failed drive, but want to get the procedure outlined before i power down and replace the drive. i have tried searching the forums for a howto or for any info on replacing failed raid 5 drives, but haven't seen any good results. someone that has done this, what are the steps? i absolutely need to retain all information on the raid, losing the data would be painful at this point.

thanks for any help, it's greatly appreciated.

slate

[EmmEff]

Thankfully, it's been a few months since I've last had to do this but I think this is the syntax I use:

mdadm /dev/md0 --add /dev/hdb1

It's up to you to partition the drive first using fdisk.

The rebuild should start to occur automatically.

FWIW, I am running software RAID5 with 5 x 160GB Maxtor drives all on separate EIDE channels. Been running this configuration for almost 18 months now. I've experienced 2 drive failures during this time.

Also, refer to the The Software-RAID HOWTO for more info.

[slaterson]

thanks for this, it definitely helps and will save me some time.

to review the process... i power down, take out hdb, put a new hdb in (it is already fdisk'ed properly). power up the system and run the mdadm -add command to add the /dev/hdb1? or will the rebuild happen automatically and all i have to do is check the status to be sure it is running and completes properly?

my system has 8 ide drives. 7 of them are 300 gb, the last drive is 400 gb. i have a partition on the 400gb drive that matches the size of the complete 300 gb disk. i used seagate drives for my raid, i have been running it for a couple years now and this is the first failure i have had. regarding the maxtor drives, i bought a single 160 gb disk a few years back and it failed within the first few months of usage. maxtor replaced it for free, however i have sworn off maxtor drives since that experience.

thanks!
slate

[EmmEff]

[slaterson]

ok, so when i do the '--add' it will not create a new, empty arrary. instead, it will rebuild the existing array and all my data will be intact, as it was before the drive failed? is this your experience?

it's a big volume and it's roughly 75% full. it would really suck to lose it all.

[EmmEff]

[slaterson]

thanks for the help, emmeff! i shut down the system and put the new drive in. it's rebuilding the spare as i type this, so far so good.

thanks for the info, it helped out!

slate

[EmmEff]

Glad to hear it worked