Hello, Already done that a few years ago, it was /almost/ the same setup, except I was planning to replace two RAID1 hard drives with bigger ones, same number. What I did was : - backup all data - fail and remove disk 1 from array - replace it with new one, only bigger - wait for sync to complete - fail and remove disk 2 - replace it with new one - wait for sync to complete - grow logical volume (don't remember exactly, but you have to first grow the PV, then the VG, then the LV, that makes sense) - grow the filesystem on the LVM
I did all this without downtime but with some patience, obviously. Of course, this can be adapted to your setup, but from my point of view, it looks quite similar. Hoggins! Le 01/05/2019 à 17:28, Gregory Avedissian a écrit : > Current server setup is raid1 with three drives. (pair and a spare) > with lvm. > /var is on its own logical partition, and the library has outgrown the > disk. > > The motherboard has four sata inputs. I want to reduce the existing > raid to two drives so I can put in two large drives and make a second > raid to hold /var. > The three existing drives are sda, sdb and sdc. I want to remove sdb > and put sdc (the spare) in its place. > > The question is, what order do I do things? Which drive(s) do I fail, > and when? Can I just shut the machine down, remove sdb and plug sdc > into the second sata input? Then reboot and remove the missing sdc > from the array? Does mdadm identify the drive by its device name or by > the uuid? Will it get mad at me if I change the order of the drives? > > Thanks, > Greg > _______________________________________________ > Rivendell-dev mailing list > [email protected] > http://caspian.paravelsystems.com/mailman/listinfo/rivendell-dev >
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Rivendell-dev mailing list [email protected] http://caspian.paravelsystems.com/mailman/listinfo/rivendell-dev
