Once upon a time, Gary Stainburn gary@ringways.co.uk said:
I've used MDADM before on previous servers, but have found that this setup isn't hot swap. Ultimately if I had to replace a drive it involved a lot of effort, especially the first drive.
I use mdadm RAID in a bunch of places; it isn't automated, but it isn't that hard to replace a drive. It would be nice if the storaged project had support for this (I think the various needed bits are there, just needs somebody to write a front-end to make the calls I guess).
By using H/W RAID, it's literally just a case of removing the dead drive and inserting the replacement. I've got a number of IBM and DELL boxes like this. it's just a pity they're not compatible with Linux so I can't monitor or manage them while the servers are running. The only way I know I have problems is by watching the LEDS
I also use Dell servers with various hardware RAID cards in a bunch of places. I install the Dell tools (they have yum repos for this), and then can use omreport and omconfig to monitor and manage RAID from within Linux, and their SNMP agent to monitor system health from my external monitoring system.
One nice thing with omreport/omconfig is that it doesn't matter what type of RAID card/chip/whatever they use (because they change things over generations), the commands are the same.
Newer Dell servers also have integrated RAID management to the DRAC directly, so you can monitor/manage/etc. through the out-of-band web and SSH UI.