Greg Lehey wrote:
>
> [Format recovered--see http://www.lemis.com/email/email-format.html]
>
> On Saturday, 18 March 2000 at 3:34:38 +0100, Palle Girgensohn wrote:
> > Hi!
>
> Please don't send messages one line per paragraph. It's a pain to
> reformat.
Yeah, I had had fiddled with the setting for a specific purpose,
but forgot to set them back. Sorry about that!
> > Following the instructions in UPDATING, when rebooting to single
> > user mode, vinum wouldn't work since the kernel module was out of
> > date - no surprise.
>
> Hmm. After updating, you should have had new klds as well. But
> that's probably not your fault. Could you enter a PR about it,
> please?
Yes. It's a documentation bug, as has been pointed out by Daniel
C. Sobral.
> > So, I copied a fresh vinum.ko in there and tried again. This time,
> > vinum loaded fine, but complained that it couldn't get the list from
> > disk (or similiar).
>
> How similar? That statement doesn't really help very much. Vinum
> produces error messages to help pinpoint problems.
Unfortunately, I didn't write it down. I regret it. Here's
briefly what happened: since 'start' didn't work, I tried to read
the configurations off the disks one by one, which wasn't a very
good idea, apparently, for since they weren't all started at
once(?), some plexes were marked a faulty. I rebooted the
kernel-3.x and started vinum with the old kld. It started and
read all disks, but some were still marked faulty or flaky. I
stopped and restarted the plexes/disks/subdisks quite a bit
before got them all up again. It seems to me that vinum sometimes
isn't quite logical about its decisions as to whether a
disk/plex/sd is up or flaky. Is there a trick with the restarting
sequence if a disk is marked flaky? I got the error 16(?) (device
busy) a lot, and had to reboot again to get rid of them.
After installing all klds and remaking the devices, I got the
kernel-4.0 to read all the disks with the start command.
> > 3-stable kernel, make first installed the make binary itself, and
OK. Here's another documentation bug, imho. I missed moving the
/etc/rc.conf.local away. make all depends on target
upgrade_checks and it installs make if the test target fails. I
think there should be a note to run make test before
installworld. My make binary was blown away with no warning, and
I was lucky to have another 3.x system left to fetch it from...
> It looks like you shot yourself in the foot.
Yeah, that's one way to put it :)
> I'd have to find out what went wrong first. It looks as if it should
> have worked modulo the problems installing the klds.
Yep. I'm preparing a pr for documentation bugs.
Thanks for your time.
/Palle
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
