Re: [zfs-discuss] Re: Re: How to destroy a pool wich you can't import
On Thu, Sep 07, 2006 at 06:31:30PM -0700, Darren Dunham wrote: > > It certainly changes some semantics... > > In a UFS/VxVM world, I still have filesystems referenced in /etc/vfstab. > I might expect (although have seen counterexamples), that if my VxVM > group doesn't autoimport, then obviously my filesystems don't mount, and > that will halt startup until I deal with the problem. This is often a > good thing. > > With ZFS and non-legacy mounts, I don't really have a statement that the > ZFS filesystem /path/to/critical/resource must be mounted at boot time > other than the configuration of the pool. I guess I need to make some > more explicit dependencies for services if I want some of them to > notice. (Unfortunately creating/removing dependences takes a bit more > work than maintaining a vfstab today). Currently, a faulted pool will not prevent you from coming up (that is, filesystem/* services will continue to come up). There are already some folks thinking about how failed /etc/vfstab mounts should affect boot (not everyone wants it to fail coming up). Similar thought should probably be given to what it means for a faulted pool and/or dataset. - Eric -- Eric Schrock, Solaris Kernel Development http://blogs.sun.com/eschrock ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Re: Re: How to destroy a pool wich you can't import
On Thu, Sep 07, 2006 at 06:07:40PM -0700, Anton B. Rang wrote: > > And why would we want a pool imported on another host, or not marked > as belonging to this host, to show up as faulted? That seems an odd > use of the word. Unavailable, perhaps, but not faulted. > That's FMA terminology, and besides wanting to stay within the same framework, I believe it is correct. If you have booted a machines which claims to be the owner of a pool, only to find that it has since been actively opened on another host, this is administrator misconfiguration. As such, the pool is faulted, and an FMA message explaining what has happened, along with a link to a more detailed knowledge article explaining how to fix it, will be generated. The term 'faulted' is specific FMA terminology, and carries with it many desired semantics (such as showing up in the FMA resource cache). Silently ignoring failure in this case is not an option. If you want this silent behavior, you should be using some combination of clustering software to provide higher level abstractions of ownership besides 'zpool.cache'. - Eric -- Eric Schrock, Solaris Kernel Development http://blogs.sun.com/eschrock ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] Re: Re: How to destroy a pool wich you can't import
> And why would we want a pool imported on another host, or not marked > as belonging to this host, to show up as faulted? That seems an odd > use of the word. Unavailable, perhaps, but not faulted. It certainly changes some semantics... In a UFS/VxVM world, I still have filesystems referenced in /etc/vfstab. I might expect (although have seen counterexamples), that if my VxVM group doesn't autoimport, then obviously my filesystems don't mount, and that will halt startup until I deal with the problem. This is often a good thing. With ZFS and non-legacy mounts, I don't really have a statement that the ZFS filesystem /path/to/critical/resource must be mounted at boot time other than the configuration of the pool. I guess I need to make some more explicit dependencies for services if I want some of them to notice. (Unfortunately creating/removing dependences takes a bit more work than maintaining a vfstab today). -- Darren Dunham [EMAIL PROTECTED] Senior Technical Consultant TAOShttp://www.taos.com/ Got some Dr Pepper? San Francisco, CA bay area < This line left intentionally blank to confuse you. > ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Re: Re: How to destroy a pool wich you can't import
A determined administrator can always get around any checks and cause problems. We should do our very best to prevent data loss, though! This case is particularly bad since simply booting a machine can permanently damage the pool. And why would we want a pool imported on another host, or not marked as belonging to this host, to show up as faulted? That seems an odd use of the word. Unavailable, perhaps, but not faulted. This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] Re: Re: How to destroy a pool wich you can't import
This could still corrupt the pool. Probably the customer has to write its own tool to import a pool using libzfs and not creating zpool.cache. Eventually just after pool is imported remove zpool.cache - I'm not sure but it should work. This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss