|
|||||||
This is a discussion on Calculation of "Free Space" for snapshot auto-deletion within the RollBack Rx forums, part of the Disaster Recovery Programs category; Howdy folks! I have a question about one of the "Advanced Settings" in the RollbackRx console, under Program Settings. I ...
![]() |
|
|
LinkBack | Thread Tools | Display Modes |
|
|||
|
Howdy folks!
I have a question about one of the "Advanced Settings" in the RollbackRx console, under Program Settings. I can enable the checkbox for "Delete the oldest unlocked snapshot when free space is below x MB," but how is this free space calculated? I have a pretty complicated setup, I'm using BootIt NG to specify different boot configurations. This means each time I boot, there are different partitions visible to windows. Some partitions which are protected under Rollback are hidden from windows. All of my Rollback-ing is done through scripts, none of it through the console. So, when a new snapshot is being created, does Rollback calculate the free space for EACH PARTITION, or only for the OVERALL SYSTEM? There may be 20 gigs of free space for the whole system (across all my protected partitions), but if there's only half a gig free on any one partition, I want Rollback to delete some snapshots until I have enough space on that partition. According to this post, data for a snapshot is only stored on the partition on which the changes have been made, is that right? I will probably have many more snapshots on one partition than on any other (they won't be evenly distributed). If necessary, I can write some scripts to calculate free space and delete snapshots which take up the most space on whichever partition is running out of space, but that would be particularly nasty. I'd much rather click a checkbox. This would be really painful for me to test how it works, so I was hoping that someone on this forum might have figured this out for themselves, or that a developer might drop and explain it to me. |
|
|||
|
Quote:
Welcome to the Horizon DataSys forum. For reference, support can be reached at the Horizon DataSys Facebook page Horizon DataSys | Facebook. Click on "Support" on left-hand side and then click on the "Support Request" Tab. Support can also be reached at http://support.horizondatasys.com and let's see if we can get some feedback here. Best, Jacob |
|
|||
|
Quote:
Thanks? |
|
|||
|
Sounds very hair-raising to me. We don't know a lot about how RBRx works inside, we do know it keeps a database to track the sector substitutions, but not where the database is kept in detail so any jiggery-pokery of hiding partitions in multiple boot configurations is asking for trouble.
One thing you really must avoid is altering any partition that RBRx is supposed to protect while RBRx isn't running. However, by what you say it seems you are having some success, and it would be good to know more in detail because it will enable us to surmise more about the inner workings. I only have RBRx protecting one partition, so naturally in my case the calculated free space applies to that partition. |
|
|||
|
It seems nobody else has my specific needs for Rollback, which is fine: I realize I'm asking more of it than the typical user. I ended up doing my own testing to find out the answer.
Quote:
* Rolled back to baseline * Made ~1 gig worth of changes (different changes those on the previous snapshots) * Took a snapshot Rollback was set to delete snapshots when there was less than 1 gig free space, and no snapshots ever got deleted. So, free space is calculated for all free partitions (as indicated in the Rollback GUI console), and snapshots are in fact distributed across all protected partitions. I had ~ 5 gigs worth of snapshots taken, and only 2 gigs of free space on the partition on which the changes were happening. So, I ended up making my own tracking system for snapshots and wrappers for the shieldcmd.exe function calls so I only store an arbitrary number of snapshots at a time. Rollback still works great, it just doesn't have specific enough snapshot auto-delete functions for what I'm trying to do. This is just a heads up to any other other poor soul who may stumble across this forum in search of an answer to a similar question. |
|
|||
|
Hi soccerdog8. Sorry i didnt reply earlier but i only just noticed that i had a PM from you.
![]() I've read all of this thread and i'm not entirely convinced that your conclusions are right. Your test was very interesting and quite convincing but i need you to do something for me. Hopefully you still have this experiment set up. I would like you to look at the snapshot tree of the 5 snapshots and take note of the snapshot sizes. Presumably they would be around 1Gb in size. Do a Snapshot defrag and take a look at the snapshot tree again. Are the sizes still 1Gb each? The reason i ask this is because as we know, Rollback works on the sector level. The power of Rollback is its ability to "share" common sectors between the snapshots and so the resulting snapshots need only be as large as the "change in data" from the previous snapshot. This "change in data" resides on the drive it is originally created on or resides on the drive you've selected to put it on. I have never seen any evidence to suggest that Rollback would split a file's sectors that i want on say drive D: spread over onto drive C:, until your little experiment came along. It gives me pause. However, i'm hoping that after you run the internal Rollback defrag that some more light would be shed. If i may jump the gun, i just dont think that that is how Horizon would design Rollback because to me, splitting files across partitions would make for very slow reads times would it not? |
|
|||
|
Alright Carfal, you win
![]() I decided to run this test again but set it up more intelligently. This time I protected two partitions. Windows XP, with 1.7Gb free space; and Windows Vista, with 15Gb free space. Then I set Rollback to delete snapshots when there was only 5000Mb free space left and I set to work taking snapshots on WinXP. I let my little system run for a while, having set it up to take a maximum of 20 snapshots, knowing full well that there wan't enough space on the WinXP partition for that many partitions. So you can see the dilemma: was Rollback going to start deleting snapshot immediately because there's less than 5Gb free space on the WinXP partition? Was it going to take 12Gb worth of snapshots on WinXP before it started deleting snapshots? (If so, that would prove that snapshots are distributed across partitions pretty clearly) Or was it going to take 2Gb worth of snapshots on WinXP and die some horrible death because it was trying to save snapshots on a full partition? (This would be a pretty bad bug, in my opinion) Next time I checked, my machine was stuck in the Rollback console giving me some nasty looking errors. It was failing to restore to the baseline snapshot. ![]() ![]() There were several of these errors, maybe 3 or 4 of each type. Each time, if I clicked OK, it tried rolling back to the baseline snapshot again and gave me the same error. If I clicked cancel, it brought me to another error, then another error, until finally the snapshot restore finished and I got to my desktop. ... and I'm out of image allocations for this post. Continuing in the next post ... |
![]() |
| Thread Tools | |
| Display Modes | |
|
|