Other setting up raid 1??

[tony33]

I own a server and have one hard drive for years. I installed freebsd 8.1 on it and have updated it to the current version. It's been years I have used this drive. However, recently bought 6 hard drives and want to now use the raid options. however, when I tried using the built in raid controller. I was notified that the hard drives will be wipe cleaned. Is there any way I can setup a raid with old hard drive and a new drive in raid 1 without having to wipe both clean. raid 1 is mirroring and I don't get why it needs to wipe the original drive clean. I just want to set it up to just copy the main drive and make the 2nd drive as a backup in case the drive fails.

I have 5 drives and don't know what's the best way to handle this. Should I put 2 new drives in a raid 1 and then copy my main drive over to one of the drives? Or is there a way where I can select the main drive to copy from and have the new drive be setup as the backup. So, instead of wiping my main drive clean it will just copy that drive over to the new drive and have it setup that way?

I looked online and most said that a first raid setup of any type needs both drives wiped clean. I can understand raid 0 and raid 10 but don't get why raid 1 requires this? I looked online and most suggest to have 2 new drives in a raid 1 setup and then copy over the main drive to one of the drives. I was told it should work.

Then there's others that said it depends on the raid controller use and was told Intel raid contollers allow a migration option where any raid config can convert existing drives into a raid system without data loss. Then there's people that show for a windows machine using disk management you can mirror hard drives on the fly but when seeing videos of this those are setup as dynamic drives. I have no clue if they were already setup in a raid envirement and then later config or changed the config.

I am new to raid and this would be my first attempt at it. I am just worried because the old drive has very valuable information and data on the drive that would take me years and years to redo or replace.

Right now thinking to backup the hard drive to a external drive but that external drive sometimes works meaning sometimes in the freebsd os it mounts and sometimes it fails to mount. I don't really want to be put into that situation. I might have to buy a new external hard drive. It's just that I don't want any room where I can lose this data.

I would like to know what's the best step to setup a raid 1? Do I need drivers installed for raid?

I have 5 brand new drives and 1 is old so a total of 6. should I take out the old drive. Put in 2 new drives and then setup raid 1 on them and then connect my old drive and run clonezilla or some usb bootable device that will run a program that will make an exact copy of my old drive onto one of the new drives in the new raid 1 config? Is that the best way? the old drive has freebsd os.

 

[Phishfry]

Is there any way I can setup a raid with old hard drive and a new drive in raid 1 without having to wipe both clean.

I am going to take a wild guess here and assume this is RAID on Motherboard from your old drive.
We call this SoftRAID. FreeBSD can read some software RAID arrays:
https://www.freebsd.org/doc/handbook/geom-graid.html

Your new FreeBSD RAID1 should be gmirror if you choose UFS.
https://www.freebsd.org/doc/handbook/geom-mirror.html

So you might be able to recreate the soft RAID array and transfer it to the new gmirror RAID1 array.
You could just dump/restore old softraid drive (native tools instead of clonzilla)to a new single drive and add your gmirror later.
You can add as many mirrors as you wish after the fact. See section 18.3.3 Creating a Mirror with an Existing Drive

To use your old softraid disk you really should use the original motherboard that created the array if possible.

I am just worried because the old drive has very valuable information and data on the drive that would take me years and years to redo or replace.

Well if this is the case then proceed carefully. Make sure you do a backup of the softraid disk before messing with it.

 

[ralphbsz]

I was notified that the hard drives will be wipe cleaned. Is there any way I can setup a raid with old hard drive and a new drive in raid 1 without having to wipe both clean?

Let's work through the simplest scenario: You have been using a single drive. You want to add a second drive, and you want to have all data mirrored on both drives.

The answer is: It depends on what RAID technology you are going to use, which in turn depends on how your original drive was used. Let me explain why. In order to use a disk drive in any form of RAID requires storing a little bit of metadata. The RAID implementation has to write down, somewhere, that drive X is not an individual disk that can be exposed directly to the operating system as a block device, but that it is a part of a RAID group. How complex that metadata is depends on the RAID implementation; the simplest is probably a simple 2-disk RAID 1 setup with identical-size drives, where we really only need to say "disk X (which is vendor Xv, model Xm, revision Xr and serial number Xs) is one half of a pair with disk Y (=Yv, Ym, Yr, Ys). That metadata could be stored in a single 512-byte block. For more complex RAID implementations, the metadata that describes the mapping form physical disks to virtual arrays can be much more complex and much more voluminous (I've worked on RAID implementations where the metadata is many GB in size). Regardless how complex the metadata is, it needs to be stored somewhere. The best place for it is on the disk drive itself (perhaps with more copies elsewhere, but there are strong technical reasons that the RAID metadata be on the drive). That means that the "useful" capacity of the drive will be slightly smaller, since the RAID implementation has to steal a little bit of space for the RAID metadata.

This is one fundamental reason that a non-RAIDed drive can not always be turned into a member of a RAID set: If we already formatted a file system on the disk, but then have to reduce its capacity, that won't work, without breaking the file system.

The other fundamental reason is this: that RAID metadata has to be in an easy-to-find place. Many simple (hardware) RAID implementations simply put a descriptor area at the beginning of the drive, and shift the real content of the drive over. Clearly, this can't be done if there is already useful file system data at the beginning of the drive.

Some RAID implementations can deal with this. One example is ZFS (which does include a RAID implementation): In its ZFS-internal metadata, it always leaves a little space for this, and it can always turn a non-mirrored pool into a mirrored pool. I know that some hardware RAID implementation (I think some of the LSI logic cards) can import a non-mirrored disk into a mirror pair, but it takes a few hours (I think they do this by copying the data once to shift it away from the beginning of the disk, reducing the capacity, and hoping for the best).

In a nutshell: you are asking for something that may be impossible.

BUT: Here is good news. You used to have 1 drive, you now have 6. Why don't you create a sensible RAID set from the 5 new ones (they are right now empty), copy the data form the old drive onto them, then recycle the old drive into the RAID set?

Another question: What type of RAID are you planning to use with your new 6 drives?

 

[tony33]

Let's work through the simplest scenario: You have been using a single drive. You want to add a second drive, and you want to have all data mirrored on both drives.

The answer is: It depends on what RAID technology you are going to use, which in turn depends on how your original drive was used. Let me explain why. In order to use a disk drive in any form of RAID requires storing a little bit of metadata. The RAID implementation has to write down, somewhere, that drive X is not an individual disk that can be exposed directly to the operating system as a block device, but that it is a part of a RAID group. How complex that metadata is depends on the RAID implementation; the simplest is probably a simple 2-disk RAID 1 setup with identical-size drives, where we really only need to say "disk X (which is vendor Xv, model Xm, revision Xr and serial number Xs) is one half of a pair with disk Y (=Yv, Ym, Yr, Ys). That metadata could be stored in a single 512-byte block. For more complex RAID implementations, the metadata that describes the mapping form physical disks to virtual arrays can be much more complex and much more voluminous (I've worked on RAID implementations where the metadata is many GB in size). Regardless how complex the metadata is, it needs to be stored somewhere. The best place for it is on the disk drive itself (perhaps with more copies elsewhere, but there are strong technical reasons that the RAID metadata be on the drive). That means that the "useful" capacity of the drive will be slightly smaller, since the RAID implementation has to steal a little bit of space for the RAID metadata.

This is one fundamental reason that a non-RAIDed drive can not always be turned into a member of a RAID set: If we already formatted a file system on the disk, but then have to reduce its capacity, that won't work, without breaking the file system.

The other fundamental reason is this: that RAID metadata has to be in an easy-to-find place. Many simple (hardware) RAID implementations simply put a descriptor area at the beginning of the drive, and shift the real content of the drive over. Clearly, this can't be done if there is already useful file system data at the beginning of the drive.

Some RAID implementations can deal with this. One example is ZFS (which does include a RAID implementation): In its ZFS-internal metadata, it always leaves a little space for this, and it can always turn a non-mirrored pool into a mirrored pool. I know that some hardware RAID implementation (I think some of the LSI logic cards) can import a non-mirrored disk into a mirror pair, but it takes a few hours (I think they do this by copying the data once to shift it away from the beginning of the disk, reducing the capacity, and hoping for the best).

In a nutshell: you are asking for something that may be impossible.

BUT: Here is good news. You used to have 1 drive, you now have 6. Why don't you create a sensible RAID set from the 5 new ones (they are right now empty), copy the data form the old drive onto them, then recycle the old drive into the RAID set?

Another question: What type of RAID are you planning to use with your new 6 drives?

The reason is I am new to raid and don't want to mess things up.

I am thinking to buy 1 more hard drive to replace the old drive. I have 5 identical drives got a good deal on them. the 1 that is old isn't identical. I am thinking to buy another drive that is the same from the same supplier. The deal is still going on. I will wait to get the drive.

I am just worried that the process to transfer from the good working hard drive that data can always get corrupted.

I have used dump before to make backups of partitions. However, never actually did a restore and still not sure if I am doing the dumps correct. However, talking with people and showing commands used they tell me that I made a full working backup. Or done it correctly.

I right now cannot afford doing a a dump and restore. Do you think I should practice ? like should I connect a new drive and like do a restore. Take my old drive out and put only my new drive and see if it boots?

I just want to make sure I can practice and learn how to do this properly.

Now about the raid 1 setup. If I put 5 new drives and set lets say 2 up for a raid 1. would I use dump to restore? what do you exactly mean to transfer the files tot he new raid array?

Like should I use dump or is there something else I can use? the old drive isn't identical to the new drives? The new drives has more storage space than the old drive.

the raid controller is on the actual server motherboard and it's made by supermicro.

 

[chrbr]

I right now cannot afford doing a a dump and restore. Do you think I should practice ? like should I connect a new drive and like do a restore. Take my old drive out and put only my new drive and see if it boots?

I just want to make sure I can practice and learn how to do this properly.

It is worth to practice that. Otherwise you backups are less useful. But there is no need to start on your main system. You could attach a disk or thumb drive and add a file system with some content. Then you can try to restore that on a different disk or thumb drive. If that works you are sure that the first step is done. Basically it is a good practice to verify that restoring backups is successful from time to time. Getting practice with some test drives and test data does not harm in case of mistakes, but it gives confidence if it works well.