Date: Wed, 15 Sep 1999 21:47:19 -0500 From: "Jeffrey J. Mountin" <jeff-ml@mountin.net> To: Michael Robinson <robinson@netrinsics.com> Cc: freebsd-stable@FreeBSD.ORG Subject: Re: Vinum performance testing... Message-ID: <3.0.3.32.19990915214719.01d2e770@207.227.119.2> In-Reply-To: <199909160143.JAA18180@netrinsics.com> References: <3.0.3.32.19990915193638.01de52d0@207.227.119.2>
next in thread | previous in thread | raw e-mail | index | archive | help
At 09:43 AM 9/16/99 +0800, Michael Robinson wrote: >My biggest objection to vinum is that it will do what you tell it to, >and then complain afterwards, rather than complain about it at the time >you try to do it. Two good examples: > >1. You can add drives partitions that haven't yet been MAKEDEV'ed > into the filesystem. They will be added to the vinum configuration > information. Then vinum will give profuse messages about > inaccessible devices. Not quite sure I follow this. I *may* have seen similar behaviour in the past, but think of it as pilot error that was easy to fix. Were you using sysinstall to setup the partitions? I find that post-install use of sysinstall doesn't always create the devices. >2. You can create a striped plex from unequally-sized subdisks. The > plex will be added to the vinum configuration information. Then > vinum will give profuse messages complaining that striped subdisks have > to be the same size. Pretty sure that this is not advised and you should ensure that striped plexes are the same size. Never tried using a length of zero on uneven subdisks (should work, IMO), but see your point. Looks like some sanity checks are needed to avoid pilot error. >This was 3.3-RC bits, last week. The specific action was this: I had >a 0+1 configuration. One of the mirrored plexes was suffering from problem >#2 above, so I had deleted it. However, because the "vinum start" command >apparently opens all the partitions and keeps them open, I couldn't change >the disklabel on the odd-sized partition. So, I tried to manually start >the other plex by "vinum read disk1" followed by "vinum read disk2", >instead of "vinum read disk1 disk2". What happened after that was the >two subdisks of the striped plex were essentially trashed. All sorts of >really cool kernel side effects were visible at that time, such as streams >of garbage characters on the console that precluded an orderly shutdown. >A couple of hard restarts later, and there wasn't anything left to salvage. Not up to trying this at the moment, but you should have been able to fix this. Did so with a 0+1 setup, but it was a bit tedious. Don't care to answer this one publicly, since what I do may not be the "proper" way, but it does work and allows some variation. Documentation or examples pertaining to fixing and recovering from problems is lacking, as well as adding/removing objects. The latter I pointed out to Greg after mucking my way through it. >I have extensive experience with Veritas Volume Manager under Solaris. In >many ways, I think vinum is an excellent replacement. However, it very much >seems that it is still in the "try to make it work" stage of development >and hasn't yet entered the "try to break it" stage. One thing mentioned quite a while back the configuration management need work and your experiences show that it still does. Jeff Mountin - jeff@mountin.net Systems/Network Administrator FreeBSD - the power to serve '86 Yamaha MaxiumX (not FBSD powered) To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3.0.3.32.19990915214719.01d2e770>