gjournal volume remount after os clean reinstall

S

silkie

Hi all, my first post here but I have used FreeBSD for years at home. My knowledge isn't great with FreeBSD.

I need to understand how to remount a jorunaled volume after an clean os reinstall.

I have a 7.0 install with a sperate journaled disk. (data and metadata on that same disk) The data content is valuable so I'm little nervous.

My question is what is the correct procedure for remounting after a clean install. (incidentaly installing 7.2-RELEASE)

Can I just mount the data slice of the volume as a normal UFS mount or should I rebuild the journal slice in the new OS and mount as a jounraled volume and with either of these options is there any danger to loosing the data?

P.S. I can't backup the data on the gjournaled volume as the machine hangs during the copy. :r The reinstall to 7.2 will be introducing a new mobo and cpu etc so should stop the random hanging!
 
After a clean install, you should be able to reattach the separate drive and load the geom_journal.ko kernel module, after which the journaled volume should be detected correctly, i.e. some drive.journal device node appears in /dev, which you can mount again.

I'm not aware of the journal needing any rebuild between FreeBSD upgrades, it should be pretty self-contained.
 
To summarize, I think that all that needs to be done after a reinstall is:

- reattach the journaled drive
- load geom_journal.ko (e.g. manually first to test, and later through /boot/loader.conf)
- mount the /dev/whatever.journal like you did before, maybe later readd it to /etc/fstab

I wouldn't mount the raw data partition without the geom_journal.ko module active... I don't see the point of that.
 
Thanks......

Sfynx - cheers matey.

Pretty much what I thought would be the case but I was only assuming and you know what people say about that!

As for mounting the data slice direct, let me explain my rationale.

When building the journal you obviously have to explicetly mention the dev node of the data and metadata providers. i.e. /dev/ad6s1b and d in my case.

It just so happened I had to boot without ad6 connected and the gjournal module (or maybe it was the duff fstab entry) that noticed ad6 wasn't connected, aborted the boot and slung me into single user mode.

Obviously when I put my nice new HW in I can't garranty that disk is going to be on ad6. So On boot will the gjournal module regonise the providers as ad6 and as the disk is now on ad4 (for example) then stop the boot? Far too many unkowns for me, so I thought the easy option would be to mount the data direct. Copy it to a safe volume and then mess around rebuilding the jounral on the new dev nodes.

However while reading this back I assume only a duff fstab entry is going to kill the boot process..... maybe all my worry is for nothing! I'm sure the gjournal devs have though about this scenario and the device node refferances will all be paramaterised.

Wish me luck with my hardware swap and thanks for the clean re-mount instrcutions.
 
Sorted

The gjournal volume was fine with the new device node in the new system.... conculsion: Have more faith in the FreeBSD coders. :P
 
You must log in or register to reply here.
Back
Top