drives, daily backup from drive 1 to drive 2 using rdist-backup. Data is > written only to 1 drive. It has 4GB RAM. Small subnet - 3 fixed NFS clients
Maybe our config/setup is helpful for ideas, now or if you grow : We use mdadm raid1 for all our boxes, NFS servers included. It buys days of time after losing a disk before we have to replace it. *Rdiff*-backup is used daily to another box, for box/site disaster and hist archiving. If intended for datacenter, then we now prefer Supermicro 1U small servers. Previously we did use the cheapest Lenovo's but they would get wedged sometimes, and require travel to DC to reboot. Supermicro is no-frills, has IPMI for remote mgmt, and worked well for us for years. Downside is no hotplug, so have to bring box down for few minutes to replace bad drive. 2x2.5 drives fit easily on the chassis we chose. We use NFSv4 to export spare disk from all boxes to all boxes, with home dirs local to each box. Users are asked to keep dot files in home dir or any files they do not need backed up. Each is given work dir on nfs, visible on all boxes - these are backed up by replicating periodically. All network dirs are accessed through symlink, so recovery is manually updating it to the working copy. Sometimes we iscsi-export partition from diskrich box to diskpoor box needing faster local disk than NFS can do. Maint work is rare, triggered when hourly monitoring scripts report failed / nearly full disks. OS is Centos7 as it scales to enterprise quality - minus the enterprise SAN cost :-) Tightly controlled and uniform network, so just plain local Linux security suffices. - Chris _______________________________________________ ILUGC Mailing List: http://www.ae.iitm.ac.in/mailman/listinfo/ilugc ILUGC Mailing List Guidelines: http://ilugc.in/mailinglist-guidelines
