possible zfs bug? lost all pools

[JoaoBR]

On Sunday 18 May 2008 12:39:11 Jeremy Chadwick wrote:
> On Sun, May 18, 2008 at 12:20:33PM -0300, JoaoBR wrote:
> > On Sunday 18 May 2008 11:11:38 Greg Byshenk wrote:
> > > On Sun, May 18, 2008 at 09:56:17AM -0300, JoaoBR wrote:
> > > > after trying to mount my zfs pools in single user mode I got the
> > > > following message for each:
> > > >
> > > > May 18 09:09:36 gw kernel: ZFS: WARNING: pool 'cache1' could not be
> > > > loaded as it was last accessed by another system (host:
> > > > gw.bb1.matik.com.br hostid: 0xbefb4a0f).  See:
> > > > http://www.sun.com/msg/ZFS-8000-EY
> > > >
> > > > any zpool cmd returned nothing else as not existing zfs, seems the
> > > > zfs info on disks was gone
> > > >
> > > > to double-check I recreated them, rebooted in single user mode and
> > > > repeated the story, same thing, trying to /etc/rc.d/zfs start
> > > > returnes the above msg and pools are gone ...
> > > >
> > > > I guess this is kind of wrong
> > >
> > > I think that the problem is related to the absence of a hostid when in
> > > single-user.  Try running '/etc/rc.d/hostid start' before mouning.
> >
> > well, obviously that came to my mind after seeing the msg ...
> >
> > anyway the pools should not vanish don't you agree?
> >
> > and if necessary /etc/rc.d/zfs should start hostid or at least set
> > REQUIRE different and warn
>
> I've been in the same boat you are, and I was told the same thing.  I've
> documented the situation on my Wiki, and the necessary workarounds.
>
> http://wiki.freebsd.org/JeremyChadwick/Commonly_reported_issues
>

nice work this page, thanks

> This sort of thing needs to get hammered out before ZFS can be
> considered "usable" from a system administration perspective.  Expecting
> people to remember to run an rc.d startup script before they can use any
> of their filesystems borders on unrealistic.

yes but on the other side we know it is new stuff and sometimes the price is 
what happens to me this morning but then it also helps to make things better

anyway, a little fix to rc.d/zfs like

if [ ! "`sysctl -n kern.hostiid 2>/dev/null`" ]; then echo "zfs needs hostid 
first"; exit 0; fi

or something as precmd or first in zfs_start_main should fix this issue


talking about there are more things, I experienced still not working


swapon|off from rc.d/zfs script does not work either

not sure what it is because same part of script run as root works, adding a 
dash to #!/bin/sh does not help either, from rc.d/zfs the state returns a 
dash

do not see sense in rc.d/zfs `zfs share` since it is the default when sharenfs 
property is enabled

man page tipo tells swap -a ... not swapon -a

subcommands volini and volfini not in manual at all

man page thar zfs can not be a dump device, not sure if I understand it as 
meant but I can dump to zfs very well and fast as long as recordsize=128

but at the end, for the short time zfs is there it gives me respectable 
performance results and it is stable for me as well

-- 

João







A mensagem foi scaneada pelo sistema de e-mail e pode ser considerada segura.
Service fornecido pelo Datacenter Matik  https://datacenter.matik.com.br