From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <6256dcf20808020659r18a22e6iedc08662f89e74c2@mail.gmail.com> Date: Sat, 2 Aug 2008 09:59:08 -0400 From: "Gregory Pavelcak" To: 9fans@9fans.net MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline Subject: [9fans] Subject: Venti, redundancy, and hardware failure. Topicbox-Message-UUID: fa0221e2-ead3-11e9-9d60-3106f5b1d025 Sorry if this is a repost, but I haven't seen it show up on groups, so I think I'm OK. I recently asked about mirroring venti arenas and was told that mirroring is generally a good idea. It occurred to me that, although I know how to mirror my arenas, I'm not clear on how to recover from hardware failure. That is, suppose sdD0 and sdD1 are dedicated to arenas, they are mirrored, and venti.conf has something like 'arenas /dev/fs/arenas'. Now suppose sdD0 goes belly up. Do I change '/dev/fs/arenas' to '/dev/sdD1/arenas', in venti.conf and fossil conf? Maybe rerun fmtindex? I recall reading about setting up venti and recovering fossil from venti, but I don't remember seeing anything about this. Also, I was looking at venti-backup. I assume that using it to 'incrementally back up a Venti server to another Venti server' could also include just copying the arenas to another device on the same machine. Is that right? Even if I can use venti-backup tools in this way, is mirroring preferable when it's possible? Why or why not? (Hah! Sounds like a quiz. Bad college flashback. :-)) Greg