From: irishronan on
On Aug 15, 4:23 pm, "Mark Arnold [MVP]" <m...(a)mvps.org> wrote:
> On Wed, 15 Aug 2007 14:30:15 -0000, irishronan
>
>
>
> <ronan.mcg...(a)gmail.com> wrote:
> >On Aug 15, 2:36 pm, "Mark Arnold [MVP]" <m...(a)mvps.org> wrote:
> >> On Wed, 15 Aug 2007 13:22:13 -0000, irishronan
>
> >> <ronan.mcg...(a)gmail.com> wrote:
> >> >I have followed the instructions at
> >> >http://technet.microsoft.com/en-us/library/aa997694.aspx
>
> >> >and created a RSG and added a DB to it. However when i attempt a mount
> >> >i get the following.
>
> >> >At least one of this stores database files is missing.Mounting this
> >> >store will force the creataion of an empty database. Do not take this
> >> >action if you intend to restore an earlier backup.
>
> >> >The way i have the (clustered virtual) server configured is with 3
> >> >partitions from the SAN (2 DBs and 1 Logs Partition)
>
> >> >I have only presented one of the Database snapshots to the server and
> >> >am attempting the above. Can i continue and mount /restore mailboxes
> >> >from the DB i wish to or do i need to present both DBs and the logs
> >> >partition to the machine and configure the SG accordingly!?
>
> >> >Many thanks
>
> >> >Ronan
>
> >> SnapShots?
> >> What are you using that requires you to present a snapshot up and
> >> still need the RSG?
>
> >judging by your replay am I wrong in thinking that the only way to
> >restore mailbox data is through a RSG? (if of course the mails have
> >been purged from deleted items etc...)
> >Our backup process involves us offlining the MailStores and taking SAN-
> >based snapshots of the filesystems and then back them up to tape
> >nightly.
> >The filesystem snapshot I have restored from tape to a seperate san
> >lun and then presented the lun to the server. So now I have created a
> >RSG and a new DB in it(the new 'snapshot' lun from 5 days ago).
>
> >Now when i try and mount the DB i get the aforementioned error about a
> >missing DB. why is this and is there an easierway to restore a mailbox
> >from a filesystem with the edb file on it!?
>
> >thanks
>
> >Ronan
>
> OK, this sounds like a truly ridiculous SAN or that someone has told
> you to do something plainly wrong.
> A SnapShot does not, unless you're taking an incredibly narrowly
> defined and twisted view of "offlined" take your stores offline. And
> you wouldn't take a SnapShot and restore it into a different LUN. The
> whole point of SnapShots is that the data is merely changed block
> pointers and resides on the same LUN as the running store.
>
> If you want to get an item back then you can indeed use your Exchange
> to make an RSG and then use your SAN client software on the Exchange
> Server to mount that SnapShot as a virtual LUN (giving it a drive
> letter). You can then manipulate the RSG to see the drive letter and
> then mount the "store" (which of course isn't actually the store but
> just delta's)
>
> So, exactly what is your storage product?

ok i will be slightly clearer then with regard to our backup
mechanism. my apologies for the lack of clarity before.
We use arcserve brightstore as our backup application. as a prescript
to the backup job we run a script on the exchange server which
uses MS clust service to offline the Storage groups ,Disks and
exchange information store instance.
the SAN then takes snapshots of the database luns
script then onlines the cluster again
Luns are presented / mounted on backup server
job runs

When we want to restore a mailbox we must restore from tape to a disk
(another lun) then present the disk to the server. i then run

new-storagegroup <blah> -recovery
new-mailboxdatabase -mailboxdatabaserecover <blah> -edbfilepath <path
to the new disk/file system path/*.edb with the backup edb file on it>

then when i try to mount the DB using
mount-database -identity <blah> it gives the error in the first post.
This error is confusing me and I would like some clarity on what it
means

Ronan

From: Mark Arnold [MVP] on
>ok i will be slightly clearer then with regard to our backup
>mechanism. my apologies for the lack of clarity before.
>We use arcserve brightstore as our backup application. as a prescript
>to the backup job we run a script on the exchange server which
>uses MS clust service to offline the Storage groups ,Disks and
>exchange information store instance.
>the SAN then takes snapshots of the database luns
>script then onlines the cluster again
>Luns are presented / mounted on backup server
>job runs
>
>When we want to restore a mailbox we must restore from tape to a disk
>(another lun) then present the disk to the server. i then run
>
>new-storagegroup <blah> -recovery
>new-mailboxdatabase -mailboxdatabaserecover <blah> -edbfilepath <path
>to the new disk/file system path/*.edb with the backup edb file on it>
>
>then when i try to mount the DB using
>mount-database -identity <blah> it gives the error in the first post.
>This error is confusing me and I would like some clarity on what it
>means
>
>Ronan

Your problems run far, far deeper than this. I would stand away from
the item recovery aspect for a minute. There is no need to do all of
this offlining.

However, the message simply means that your attempt to mount the store
is resulting in the process having a hissy fit and not understanding
where you have got the stores located.

ALmost certainly down to the dodgy process that your vendor has told
you about.
From: irishronan on
On Aug 15, 5:30 pm, "Mark Arnold [MVP]" <m...(a)mvps.org> wrote:
> >ok i will be slightly clearer then with regard to our backup
> >mechanism. my apologies for the lack of clarity before.
> >We use arcserve brightstore as our backup application. as a prescript
> >to the backup job we run a script on the exchange server which
> >uses MS clust service to offline the Storage groups ,Disks and
> >exchange information store instance.
> >the SAN then takes snapshots of the database luns
> >script then onlines the cluster again
> >Luns are presented / mounted on backup server
> >job runs
>
> >When we want to restore a mailbox we must restore from tape to a disk
> >(another lun) then present the disk to the server. i then run
>
> >new-storagegroup <blah> -recovery
> >new-mailboxdatabase -mailboxdatabaserecover <blah> -edbfilepath <path
> >to the new disk/file system path/*.edb with the backup edb file on it>
>
> >then when i try to mount the DB using
> >mount-database -identity <blah> it gives the error in the first post.
> >This error is confusing me and I would like some clarity on what it
> >means
>
> >Ronan
>
> Your problems run far, far deeper than this. I would stand away from
> the item recovery aspect for a minute. There is no need to do all of
> this offlining.
>
> However, the message simply means that your attempt to mount the store
> is resulting in the process having a hissy fit and not understanding
> where you have got the stores located.
>
> ALmost certainly down to the dodgy process that your vendor has told
> you about.

I am doing the offlining because i dont want exchange writing to a DB
as I take a copy of it. We investigated a few options including VSS
etc and this *appears* to be the best option for us. In saying that we
have been running exchange 2007 for about a month now and havent
tested a restore once. (were just after migrating from imap service).

>From what you say it is very plausible that either I haven't passed
the correct parameters to create-mailboxdatabase or that the snapshot
wasn't taken in a quiesced / safe state.
So is there any advice you can offer that would result in a stable/
quiesced DB such that a snapshot can be taken other that offlining the
stores?!

Ronan