Carlos E. R. wrote:
The Thursday 2008-04-03 at 03:10 +0200, Joachim Schrod wrote:
Just make sure that you really rip out the broken disk. Just last month, I was called to a customer's site because some sleep-deprived sysadmin ripped out the working disk, and not the broken disk. Usually, I recommend my technicians to just let the RAID degrade and repair it with calm, when they are well rested -- after all, they just lost the failover possibility... ;-)
I wonder why they don't put a LED on them, so that you can light it up from some software and see "that one".
The better ones have that. But they are also the more expensive ones, and often only used in data centers. With real storage systems, one adds spare disks to keep the RTO (recovery time objective, the time when failover is functional again) low. Then, exchange of the failed disk is not so urgent any more, as long as the storage box has still some spare disks left. Disk exchange is a daily task in many data centers that I know. If one has a few 100 TBs, one has disk errors every day. (Especially, as some data centers tend to prefer smaller disks. They have often better latency, and RTO time for them is smaller.) Btw, most of my customers also don't bother with RAID 5 any more. Disks got too cheap for that, they just use RAID 10. (Well, except in the compute clusters, but HPC configurations are completely different topic.) Joachim -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Joachim Schrod Email: jschrod@acm.org Roedermark, Germany -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org