|Subject:||Re: vinum in, but not in the Makefiles?|
|From:||Greg Lehey (gr...@lemis.com)|
|Date:||Sep 21, 1998 4:13:40 pm|
On Monday, 21 September 1998 at 15:29:27 +0300, pa...@polynet.lviv.ua wrote:
On Mon, 21 Sep 1998, Greg Lehey wrote:
On Thursday, 17 September 1998 at 18:27:47 -0500, Alfred Perlstein wrote:
i noticed "vinum" got added into the source tree, heard about it a few months back, LVM for freebsd sounds awesome.
however it's not being built because it's not part of the subdirs in: src/sbin/Makefile
how beta is it? or was this just an oversight?
Congratulations for bringing light into the dark room.
It wasn't an oversight. I've been travelling the last 10 days, and I didn't have time to make 100% sure that it wouldn't break a 'make world' (I still haven't; I'm currently checking the -current version out), so in order to ensure I didn't break everybody's make again, I didn't update the Makefiles. Expect them in Real Soon Now.
How would you describe the stage Vinum is on (e.g pre alpha/alpha/beta/release).
Well, since I put it in 3.0, I'd have to say it's beta. The problems it has at the momemnt seem to be more missing functionality than bugs, though I still need to get a 'make world' through (I'm running into unrelated bugs elsewhere in the tree).
Maybe short list of what vinum can provide now in semi-production environment (without data loss and often hangs and panics)?
Well, off the top of my head:
- Striped and concatenated plexes work. RAID-5 does too, but it's not in the publicly available version.
- Degraded operations work, in other words if a disk goes down, things continue to run. ccd fails under these circumstances, and needs to be reconfigured.
- Recovery (rebuilding failed plexes) works. I'm not really happy that I've fixed all possible race conditions in rebuilding while the file system is mounted, but I haven't had any failures yet.
- It's possible that a read from a newly failed drive may fail. This is a bug or a misfeature, depending on the way you look at it. It will be fixed, but since I need to retry the whole strategy, and the reads are currently done in interrupt context, it's a big one to fix.
How it is comparing to current CCD driver?
I can't see any advantage that the CCD driver still has. Vinum isn't finished, but I think it has passed CCD in every respect.
-- See complete headers for address, home page and phone numbers finger gr...@lemis.com for PGP public key
To Unsubscribe: send mail to majo...@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message