Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I am looking for recommendations to clone a Red Hat mirrored raid to a single hard drive located in the same machine. The idea is if the servers hardware ever has an issue we have a similar hardware machine ready to go. All we would have to do is pop in the cloned drive. If the servers RAID ever failed we could just switch to the single drive to maintain uptime and restore the original configuration on the spare server with a backup.

This is a restaurant and they are open 7 days a week. We do have time from 12:am to 9:00am to perform the necessary steps for a clone and we talking about under 10 Gigs of information.

There is a database on the server.

I have looked into Rsync and Clonezilla. But I am just not confident either is capable of completing the task I want. Looking for some suggestions and possibly a step by step if you could be so kind.

share|improve this question
add comment

2 Answers

I'm not sure if that is quite a disaster recovery plan; your database and drive are still fried if there's a major upset event (flood, fire, theft...)

I'd personally look at how to recover quickly from an event. You didn't say what the database was, but you can do a dump of it then copy that directory using a standard backup program like Amanda. Or rsync selected directories to another remote server or backup storage location that is offsite.

Take stock of what it would take to get that server running again. Maybe you can build another server with a similar configuration; if the only thing on that system is the database, build a "cloned" system, then do your nightly database dump, transfer it over, then have the new system restore the database as a hot spare system.

Without snapshot support taking a clone of a running system can yield inconsistencies. Especially with the database and in-flight data in memory.

Another thing to consider, if the database is the primarily important function on this, is to cluster your database server. Multiple servers being available will provide failover support. However, this depends on a few factors; what the database is, how big the restaurant (budget) is, how much administration work you want to invest...

Bottom line...personally, I'd look at doing a regular dump and transfer of the database to minimize the work involved in restoring the system to another server if need be (or a virtualized system if you can get a VM server), and use something designed to do system backups like Amanda.

share|improve this answer
    
I agree completely. There are already backups being performed and taken offsite. This would be to give as instantaneous of a recovery as possible the idea is to have a bit for bit copy in the same server for a manual fail over that would be as quick as one of us going down to the restaurant. The head admin is very adamant about having this sort of recovery. I already suggested the same method as you but the admin was less than receptive. –  Campo Jul 2 '12 at 15:14
    
I'm not a RH expert, but if it's at all modern bits of it are dynamically created at boot time so they can't be just copied bit-for-bit, and with a database server, you can't get the database state without risking corruption. Really having a hot-standby is creating a second server, configured similarly for your additional services, and have it act as a node in a clustered database, depending on DB in use. –  Bart Silverstrim Jul 2 '12 at 15:17
    
Otherwise get a maintenance contract on the server, and use proper RAID. Chances are with a good backup strategy and decent RAID (RAID 10? 1?) your failure point will be hard disks, not the controller. –  Bart Silverstrim Jul 2 '12 at 15:20
    
I guess it is convincing time. I am in complete agreement on all your points. I am trying to make something work for someone who has a vision. –  Campo Jul 2 '12 at 15:27
    
Just had a thought. This line "if the only thing on that system is the database, build a "cloned" system,cr then do your nightly database dump, transfer it over, then have the new system restore the database as a hot spare system." This is probably the best method to satisfy him. I would create a replica of the system and keep it in the same server. Every night we would do a dump of the mysql DB and then in the event of an emergency on the main drives we could boot off the spare and just restore the most recently copied DB dump. Do I have this idea right? –  Campo Jul 2 '12 at 15:46
show 3 more comments

I'm sure others are gong to give you a hard time about why your request isn't really the best idea in implement, but that is you decision.

Simply, to answer you question, you are likely going to have difficulties accessing open file to copy them across at night. My suggestion would be to use LVM across your virtual FS on the RAID 1 and the extra drive in side the case.

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.