nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Best Practices - backing up TeamSite

Keywil

TeamSite 6.7.2 SP2, RHEL5.3

Hey all,

We have our TeamSite installation within a VM running RHEL 5.3. We employ MetaTagger (on a separate vm) and DataDeploy to record extended attributes to an Oracle database (on a third vm).

I'm trying to determine the best practice for backing up TeamSite and specifically the mount (/iwmnt), the database, and the MetaTagger/conf/ directory.

The current solution is to take nightly snapshots of all the VM's and keep those snapshots for a period of time. If there was a problem we could revert to an earlier snapshot.

My question is to inquire as to the best practice of backing up the TeamSite and MetaTagger data. The TeamSite documentation explains freezing and thawing and third party applications. As I understand things, installing TeamSite within a VM is a relatively recently supported feature of Interwoven so I thought I'd raise the question here.

Are there any pros or cons that we should be aware of with the current backup policy VS the freeze/backup iwmnt/thaw method?

I appreciate any feedback and/or advice. Thank you.

Find more posts tagged with

Comments

nipper

This is alll IMHO:

Best practices were written in 2000 when PCs had small disks and virtual machines were virtually non-existant. Yes if you are backing to tape/CD/etc and perform incrementals, you need to freeze backup, thaw.

Backing up a VM is a great approach, esp for hte guy who sells you disk Smiley Very Happy

Seriously, backup a VM. That is by far the best way. Restore TS from a backup (or several incrementals) and if you do not shoot someone, you will understand why VM is the way to go.

Crsb

This is alll IMHO:
Seriously, backup a VM. That is by far the best way. Restore TS from a backup (or several incrementals) and if you do not shoot someone, you will understand why VM is the way to go.

taking backups/snapshots of the VMs is definitely the best way when you want to restore the entire TS server, but what if you just want to restore the data/iw-store because you don't want to loose new configurations/custom code...?
is there way to take separate snapshots for specific areas in the vm? like backup for the iw-home, backup for iw-store...if not, then restoring just iw-store from a snapshot wouldn't be as easy as restoring from traditional backups.

nipper

taking backups/snapshots of the VMs is definitely the best way when you want to restore the entire TS server, but what if you just want to restore the data/iw-store because you don't want to loose new configurations/custom code...?
is there way to take separate snapshots for specific areas in the vm? like backup for the iw-home, backup for iw-store...if not, then restoring just iw-store from a snapshot wouldn't be as easy as restoring from traditional backups.

You can still do this faster with a VM than with a stand alone backup. You can load the snapshot in a temp area, start the VM, grab whatever files you need and delete the temp VM.

Migrateduser

We're on a Solaris machine (TS 6.7.1) and run a full nightly tape backup of the repository and an incremental of the rest of the server's content.

We run a pre-backup script to freeze the repository for 10 minutes, take a Solaris 'snapshot', mount the 'snapshot' to a /backup dir, unfreeze the repository and back up the /backup snapshot dir.

When the backup's done, we run a post-backup script to release the snapshot.

We've been doing this for several years and, lo and behold, this server had a serious crash last month which corrupted a store within the repository. We recovered it from tape and all's well.

Srinivasa Talla

We are talking about 2 different things : 1. code back up - like iwhome(or selected directories) 2. store back up - iwmount. When we are taking snapshot(on windows vmware), it is taking snapshot of entire system. Not sure about how we could snapshot only iwhome or iwmount. Restoring back is very simple and will take just few minutes.

Our network guys are saying we can't have multiple snapshots for more days, so we have to keep 1 or 2 snapshots and also for few days(keep 7 snapshots and delete the 1 old and add 1 new - will it be good?). Because they are saying the snapshot space is accumulating and becoming large area. Then the system performance will be slow.

The other way they are talking about is imaging the entire system and keep as separate copy on VM. But not sure making 7(or few) images will give any trouble to the VM Space/memory.

We have some more time to keep the production system on VM Ware. How you guys are handling backup with VMWare including incremental backup?

nipper

Having a copy of a VM is quite nice, esp if someone was, hypothetically speaking of course, type sudo rm -rf /etc instead of sudo rm -rf /extra

Of course if one had done that just after making a copy of their VM, say in preparation of upgrading to 7.1, then recovering takes about 30 minutes.

Of course, this is pure speculation, not that any of this really happened. Smiley Very Happy

vpatel

Of course, this is pure speculation, not that any of this really happened.

Sure about that? I just heard otherwise -- of course through some social networking. Smiley Very Happy

Keywil

Thanks for your feedback folks. It pretty much mirrored what I was thinking but it's good to know that others are of the same view.

the most important piece of this is the data itself. The TeamSite application can get rebuilt fairly quickly but the data... one of the objections that I've heard was.. " well what if the /iwmnt/ is a symbolic link to a SAN disc.. then what??"

Well I presume you'd backup that disc at the same time that you take a snap shot of the vm (IMHO)

Would it be necessary to freeze teamsite during this whoel process - probably wouldn't hurt but honestly this isn't my specialty.. any thoughts on that?

freeze >> snapshot of VM/copy SAN/copy database/copy Metatagger >> thaw