first how are you every body I'm just subscribed in your forum and i hope i found what i searched for along time .
I'm not a Solaris specialist but i read more to build a Network Management Station depends on Solaris as OS and it is working good now .
my problem is how to perform full backup and restore what i backed up .
for Backup i do the following :
#/usr/sbin/ufsdump 0ufc /dev/rmt/0n /
#/usr/sbin/ufsdump 0ufc /dev/rmt/0n /opt
#/usr/sbin/ufsdump 0ufc /dev/rmt/0n /usr
and i read for ufsrestore but could any one till me the exact required steps to restore the above mentioned files .
thanks for every one who care my question but as i told before that
1- i'm not solaris expert so i couldn't change my way and use another way to backup and restore .
2- the backup already has been taken using the ufsdump and i need to restore it using the appropriate way which is ufsrestore
so i hope to find someone who help me in my exact problem
for ufsrestore you need a running system... ufsdump and ufsrestore ist not a good way for "bare metal restore". flash archive is! cause you are no expert listen to the advices given...
zfs with snapshots is nice for data volumes but (for the lack of zfs boot) not for the root volume!
for installing the dump on "bare metal" you need a boot server to via network and a whole lot of knowledge. for a flasharchive you just need the archive and a solaris dvd to boot from!
You should. Restoring from your backup tape might break your system. ufsdump is designed to be run on unmounted (or snapshot) partition, which doesn't seem the case here.
You do not give enough details about the system where the restore should be done.
Is it the same machine than the one where the backup was built ?
Why do you need to restore ? Is the O/S broken ?
First : i'd like to thank every one for his care
second : forgive my language
i will list you all details to help me if you can as i hope :
as i told before that in my company i was responsible for installing NMS ( Network Management Station ) called Naviscore which needed to configure, monitor, and control a Lucent switch network so to build such station you need the following Software:
� Solaris 2.6 and Common Desktop Environment
� Sybase 11.9.2 SQL Server
� Open Client 11.1.1
� HP OpenView 5.01
� NavisCore 05
as you see that the system is too large and it takes too time to work . there is no problems till now but the problem located in what we built in such system simply we have 50 lucent device managed ,configured every thing you can imagine will be controlled via our Server also you can note from Software required >> sybase which mean that there exist huge data will be stored . simply all what others built through more than 8 years .
so there is a spare Machine which i use in my work and built a demo version of our work then backed up the full system ( with the same way the current server work ) and i'm trying to Restore it .
so we can rescue our current system easily and in fast way as possible we can .
i already tried the Restore command which is " ufsrestore xvf /dev/rmt/0n /opt " but i faced more errors as i don't know what he asked me for as example of errors messages :
read error while skipping over inode4 unless you know which volume your file are on you should start with the last volume and network to wards the first}.
If i but any number and press yes this last error is appear but if i press no and rewrites this command again this error is appear {media read error: I/O error}.
let me clear things more i have already server running with the above mentioned SW . i backed up the system using ufsdump then i need to restore it using ufsrestore :
1- if it success and all system recovered again with all things as i hope it will be a great thing for me .
2- if it fails as not expected i need to know how to clone the server on external media then recover the system in case any disaster occur
Basha, here is an example of the process of doing a ufsdump/ufsrestore -
Making the backup...
* Find a tape and load it into your tape device. The tape device in
this exmaple is /dev/rmt/0
* You may want to bring the system down to single user to make the
system as quiet as possible.
* Keep a record of how the partitions are set up by going into format
and copying the table to a flat file or printing it out. Just put it
somewhere that you can have access to it in the event that the disk's
partitions are totally lost. Below is an example of what you want to
record. To get to this information just su to root, type format, then pick
the disk you're backup up, then type in partition, then print.
* For even more record keeping grab the /etc/vfstab file and do a df -k
to confirm what slices are being used for what. Either print this or keep it
in a file somewhere accessible for when the server in question is not
available.
o Backup each of your separate file systems. In this case we have
/var, /usr, /opt and /. Remember or write down the order in which you do
each backup.
o ufsdump 0uf /dev/rmt/0n /
o ufsdump 0uf /dev/rmt/0n /var
o ufsdump 0uf /dev/rmt/0n /opt
o ufsdump 0uf /dev/rmt/0 /usr
o That is the extent of ufsdump. After this you will have a good
backup, so keep that tape handy.
==============================================
To do the restore ...
o Get the server in question to the OK prompt
o boot cdrom -sw ( s means single user, w means no windows )
o format ... to see if partition table is the same as it used to be.
If not, re-partition.
o Time to build a new file system on each of the partitions that need it.
o newfs /dev/rdsk/c0t0d0s0
o newfs /dev/rdsk/c0t0d0s3
o newfs /dev/rdsk/c0t0d0s4
o newfs /dev/rdsk/c0t0d0s5
o Do a files check on each of the new file systems.
o fsck /dev/dsk/c0t0d0s0
o fsck /dev/dsk/c0t0d0s3
o fsck /dev/dsk/c0t0d0s4
o fsck /dev/dsk/c0t0d0s5
o Do the actual restore, restoring one file system at a time. Make
sure you remember which backup is for which partition. For instance, if
the first file system you backed up was / , then remember to restore the
first backup to the appropriate slice. See below.
o mount /dev/dsk/c0t0d0s0 /a
o cd /a
o ufsrestore rfsvy /dev/rmt/0 1
o cd /
o umount /a
o mount /dev/dsk/c0t0d0s3 /a
o cd /a
o ufsrestore rfsvy /dev/rmt/0 2
o cd /
o umount /a
o mount /dev/dsk/c0t0d0s4 /a
o cd /a
o ufsrestore rfsvy /dev/rmt/0 3
o cd /
o umount /a
o mount /dev/dsk/c0t0d0s5 /a
o cd /a
o ufsrestore rfsvy /dev/rmt/0 4
o cd /
o umount /a
o Install the boot block.
o cd /usr/platform/sun4u/lib/fs/ufs
o installboot ./bootblk /dev/rdsk/c0t0d0s0
o The restore is now finished. If this is a test to restore a server
that is still working, you would need to change the IP of this 'test' server
or disconnect the network cable. Reboot the server and it should come
up.
Basha - that basic ufsdump/ufsrestore doesn't give you everything you may need for building out a naviscore server. Those are a very picky system to build out normally (requirement of Solaris 2.6, specific patches, Sybase, HP-OV (and it's patches)).
If you are doing mirroring with Disksuite (recommended), then you could drop side of a mirror and run your ufsdumps from it. This would give you a clean dump versus running it against a live system.
Also, I don't think you specified if you are looking at rebuilding the database server or the client(front-end) (I'm assuming you have separate servers for this) or both. As far as insuring you can recover a DB server, you should be dumping the database at the least, every hour.
Thanks again RTM i would like to tell you that DB Server is located on the same machine and i thought that when i restore the full system i will restore it also kindly could you clear this issue for me .
** from your reply i need to ask for a very vital point why Naviscore will not recovered and what is the best thing i can do to backup all things then when restore i will get all things working good again with all pre-configured data .
Naviscore is such a finicky collection of software. Just building out the front-ends sometimes takes 2 or 3 tries (with no valid reason why it failed to work on the other tries). Since you have it all on one box, I can't imagine the 'fun' you might be having. It's been around 3 years since I messed with naviscore. We had separate front-ends (3-4 each) and one DB server in production.
We used Legato backup but never restored a full server (even as a test). Restoring from Legato took so long (in those days) that it was easier to just build out another front-end. Plus, we had a Disaster Recovery site that had servers running and waiting so in a way, we were lucky enough not to have to be in a bad situation for long.
As suggested, if you are using Solaris Disksuite (aka SDS / ODS / Sun Volume Manager), then you could drop one side of the mirror, and ufsdump it to tape. Then use ufsrestore on to the same size disks on new server. I would think this would be your best solution as far as a Naviscore server is concerned.
I think you guys are not using any DisasterRecovery softwares; IMHO it is really good to have a host-based DR softwares for availing continuance for particularly these kind of critical servers; nobody can afford for losing things built through 8 years (as you said).
I can help you setting up things if you have any DR s/w.