Understanding Migration using ReadyDR to new device and preserve ALL of the current snapshots
Looking to have two NAS units which are syncronized INCLUDING snapshots:
I have been happily running my pro6 for many years, backed up a year ago and migrated to 6.6.0 and have continuously upgraded to the most recent version (waiting a couple of weeks for 6.8.0 or 6.8.1 to have some hours on the early adopters machines). The drives in this machine have LOTS of hours, they are mostly 2TB or 3TB drives, all on the list. I did recently have a drive start to develop errors and that was replaced with one of the older backup drives. System is double backed up with a rotating set of 10TB drive.
I am looking to migrate to a 626x machine and would prefer to start with all new drives (most likely 10TB seagate ironwolf drives). I would prefer to only populate 2 drives at this time and then grow as needed (no problem at this point), and then migrate the data from the old Pro6 to the 626x INCLUDING all of the old snapshots (this is where I have not been able to determine how readyDR actually works).
I realize that one path would be to install the 6 drives from the Pro6 into the 626x and then upgrade those drives to larger drives as needed (this would obviously preserve all of the snapshots) but requires having 6 drives in the system from the beginning and retaining old drives until a full 6 drives have been upgraded.
Option 2) I could use one of my current backups to load the data onto the new readyNAS but that will definitely NOT preserve the old snapshots.
Option 3) It appears that readyDR is capable working with snapshots, however, it is not clear if it works with "prior" snapshots or only the ones which are created on the source machine to perform the backup.
The bottom line is that I want two units which are syncronized, including snapshots if possible, these two devices will be physically seperated and come together once a month or so for syncronization.
Also note that these devices are not connected to the internet and have no access to one another most of the time.
Any pointers or thoughts are definitely appreciated
Re: Understanding Migration using ReadyDR to new device and preserve ALL of the current snapshots
ReadyDR won't preserve the snapshots on the original machine. It also doesn't create a duplicate (the ReadyDR backup needs to be restored before you can use it). It's closer to replicate in functionality than an NAS backup job.
You'd need to do this via ssh (using btrfs send and receive).
Re: Understanding Migration using ReadyDR to new device and preserve ALL of the current snapshots
Thank you for the pointer, I will look through the forums or on general btrfs forums outside of readyNAS.
A question about readyDR, if I restore the readyDR to a new unit do I only get the most recent version or do all of the snapshots get preserved when the readyDR volume is restored.
I am thinking about disaster recovery and trying to determine if there is a path from a readyDR backup to a system which has all of the original snapshots preserved.
Re: Understanding Migration using ReadyDR to new device and preserve ALL of the current snapshots
I don't think I understand the intent of the readyDR mechanism. It appears that readyDR is designed to backup all of the snapshots and that there is a "failback" (restore to original device) or "failover" (take over with the backup device) as described in:
What is not clear to me is if all of the snapshots are backed up (using readyDR) when I do a "failback" or "failover" do I only get the most recent snapshot or do I get all of the old snapshots?
It almost appears that using readyDR you have to select a specific snapshot to restore back to the original machine (or a single snapshot to clone in the "failover" scenerio) but that unless you go through and clone each snapshot independently only the selected snapshot will be available after the "disaster recovery".
Thank you very much for any clarification you might be able to provide.
Re: Understanding Migration using ReadyDR to new device and preserve ALL of the current snapshots
I didn't use ReadyDR for long, as it didn't fit my needs. Instead I use rsync backup jobs for each share. Unlike you, I am not particularly interested in preserving the snapshots in the backup. Instead I use the snapshots on the destination if I want to roll back to a previous backup.
My understanding is that when ReadyDR runs, it first creates a snapshot on the source and then sends that snapshot to the destination. That is incremental (only changed blocks are sent). But I don't believe it preserves all the snapshots that are on the source.
Re: Understanding Migration using ReadyDR to new device and preserve ALL of the current snapshots
Interesting...
In another comment there is an "export a seed" for the readyDR archive. I wonder what that would pull across.
Sounds like I may have to get my hands on a second readyNAS device and try some of these things and report back... Perhaps a small two bay readynas is in my future... I think it has to be x86 based or some of the features may not be available....
I think it has to be x86 based or some of the features may not be available....
Correct. Look for the RN422. The RN312 might still be on the shelves, but the RN422 is replacing it. (The RN312 is still supported of course, so if you see a good deal on it, perhaps consider it too).
Re: Understanding Migration using ReadyDR to new device and preserve ALL of the current snapshots
Well, bit the bullet and bought the 626x and a couple of 10TB IronWolf drives, planning on testing this question as I go through the process of ending up with a backup machine and a primary machine (while retaining offite backups too).
Planning on testing several of the options prior to making this new box a production server and should answer the questions I posed along the way in this link.
I went ahead and updated the new box to 6.8.0 after reading the release notes but did not notice the comment about not being able to use the UI to install apps. It notes there is a manual approach. I have uploaded apps directly in the past (primary machine never sees the internet) but still used the UI if I remember correctly. There is a comment that it will be fixed in 6.8.1, any comment on when that build will become available. This is currently not a production server so I could run a beta build and will look around at those.
There is a comment that it will be fixed in 6.8.1, any comment on when that build will become available.
Netgear generally won't say, but I think it will be quite soon. It's at release candidate 2 now, and 6.9.0 beta just started in parallel.
Thanks! I successfully "downgraded" back to 6.7.5 which is what my Pro6 is running. Just to be really sure I did not cause trouble (due to the comment about not downgrading, I performed a factory default to ensure the disk was restored to it's correct 6.7.5 status.
Just received an ST10000vnB004 drive (not on the compatability list) when I ordered an ST10000vn0004 drive (which is on the compatability list) and so the level of fun with this NAS will have to wait a few days until I can confirm that this is the same drive (perhaps a slightly different build level) or something rather different. No data on the Seagate website.
At any rate, I will plan to keep this up as documentation of my efforts to move snapshots and try using various backup methods to fully preserve the snapshots on the system, how to utilize the tools provided for full disaster recovery, and the like.
Model: RN626X | ReadyNAS 626X 6-Bay with Intel® Xeon® Quad-Core Server Processor
Re: Understanding Migration using ReadyDR to new device and preserve ALL of the current snapshots
By experimentation with ReadyDR I am beginning to understand what the behavior is when going from system to system. In particular, ReadyDR looks at the world purely from the perspective of either "shares" or "LUNs", not from the perspective of the entire filesystem. This does make sense, however, it does have some unexpected implications when trying to move to the next machine.
Specifically, if data has been moved between two "shares" and the underlying BTRFS has kept track of the original blocks the original FS may be significantly smaller than the ReadyDR share. Take the following example:
Two Shares, both initially empty (ShareA and ShareB) and a single file, FileA
Copy FileA into ShareA then take a snapshot of both ShareA and ShareB
Move FileA into ShareB and take a snapshot of both ShareA and ShareB
Move FileA back into ShareA and take a snapshot of both ShareA and ShareB
Move FileA into ShareB and take a snapshot of both ShareA and ShareB
Move FileA back into ShareA and take a snapshot of both ShareA and ShareB
Magically, the total consumed disk space is basically just the size of FileA (the wonder of BTRFS)
However, perform a ReadyDR backup of ShareA and ShareB:
The ShareA will have four copies of FileA (in different snapshots) and the total space is four times FileA
The ShareB will have three copies of FileA (in different snapshots) and the total space is three times FileA
This is all exactly what one would expect the FS to do in this case, it just means that the size of a ReadyDR backup can become quite large if there is a significant amount of moving data between shares.
I did find some experimental methods of copying an entire filesystem to new disks which might not have the same underlying structure but not knowing what the ReadyNAS is doing under the hood suggests that there might be unfortunate side effects.
If it were not for desiring to move to Raid6 I would probably just move all of the disks from one box to the other, upgrade them and then rebuild the original.
I am considering a strategy where I use ReadyDR to capture the entire share to the new machine, then clone the most recent snapshot for local use. (these machines are physically seperated and not on the internet). When I bring them together for backup I could run the ReadyDR on each share that is primarily on one box or the other. (i.e. each share has a "primary home" which is the master, it might be on either machine). After the ReadyDR jobs have completed (such that there are full backups all around) then use rsync between the cloned version and the master to incorporate any changes made in the clone, then run ReadyDR again.
The last step would be to depracate the prior clone (either delete or rename) and create a new clone from the ReadyDR archive.
For the most part, the "clones" won't get changes made to them so the frequency necessary is low.
Still working through the best way, did connect the machine to the internet to play with the other features for a bit then will reset to factory again and start working in earnest.
Re: Understanding Migration using ReadyDR to new device and preserve ALL of the current snapshots
ReadyDR behavior is exactly as expected, some of the shares are becomming quite large and there is obviously duplicates of data between the ReadyDR jobs for each individual share. For the time being "space is cheap" and so will go forward with this plan.
Currently running ReadyDR to specifically named folders on the remote NAS:
MyMusicNAS1 (on NAS1) is sent to MyMusicNAS1ReadyDR (On NAS2)
MyMusicNAS2 (On NAS2) is sent to MyMusicNAS2ReadyDR (On NAS1)
etc
There are many shares on NAS1 which don't need a counterpart on NAS2 because I would not modify the data on NAS2.
I would plan on the "MyMusicNAS2" is not the larger overall archive but rather specific versions, temporary bits, or music collected at NAS2, these are kept in a seperate share and incorporated into "MyMusicNAS1" manually.
Definitely more manual work than I had planned but safe and keeps the manual work of syncronizing data to a reasonable level. Many of these folders are largely static. If doing wholesale rearrangement it is important to do that on the original NAS.
The primary downside is being that NAS1 has 6 drives of 3TB each and all of the data would fit on a single 10TB drive (with a mirror) there is not a way to preserve the snapshot history (where it is connected in revision history) in a clone of a ReadyDR share (i.e. while the ReadyDR share has all of the snapshots included, the clone of the last snapshot in the ReadyDR share does not result in the clone having direct access to the prior snapshots.
Definitely an interesting problem (discussed on various online forums with regard to the BTRFS file system) when there are people in a larger corporation using the revision history to find things. If the primary machine fails and you recover with ReadyDR the recovered "Disaster Recovery" filesystem (i.e. a clone of the last snapshot) does not have ANY of the prior snapshots associated with it. Those snapshots ARE in the ReadyDR share and so it is possible to clone them one by one into individual shares, and it is possible if you start from the earliest and work forward that all would be linked into the same blocks inside the ReadyDR share (so the size does not expand by the number of snapshots), but the ability to walk back through the revision history is lost.
Please don't take this as a critism of the ReadyDR or BTRFS, I am just documenting the implications of how the system works because there are not many people who actually try to impliment their Disaster Recovery plan as they build up the system and want others to understand the limitations.
I am continuing to look at the ReadyNAS external enclosure and wondering if it is a seperate volume can I export it and use that to update the ReadyDR partitions on both machines. i.e. use the "export filesystem" feature for a secondary volume (i.e. /data2 rather than /data) import it into the other ReadyNAS (Yes, I know the Pro6 does not support the external enclosure) run the ReadyDR job and then move that enclosure (or drives) back to the other ReadyNAS.
