How Do You Restore Linux Files from Unitrends Image Backups
-
Apparently Unitrends takes the time to detect that it is Linux and disable the network share, just to make it not easy to restore
We already tried that route, there was nothing to connect to. -
@thecreativeone91 said:
This process doesn't work because the LVM name is a duplicate, of course. It's like they've never tried the normal scenarios and only have unusual cases. Mounting to a different machine might work, though, one that has everything mounted using different names.
-
Here is the quote from Unitrends: It is not possible to remotely mount LVM volumes on a machine that already has the same LVM UUIDs or PV IDs listed in whole or in part. Instead, select an alternate linux machine or build a new linux machine that does not use LVM mountpoints and use the iSCSI mount script from that machine and recover the files from that alternate machine. In many cases, when LVMs are in use on a Linux machine that is backed up, FLR will not be possible directly to the originating machine.
-
Recap: Unitrends doesn't support backing up and restoring to the same machine?
-
Try this from windows: http://www.diskinternals.com/linux-reader/
-
Jumped onto one of the CentOS 7 machines since its naming conventions are different. Of course, no surprise, the Unitrends mount script doesn't work there. Here is the error:
Can't locate Data/Dumper.pm in @INC (@INC contains: /usr/local/lib64/perl5 /usr/local/share/perl5 /usr/lib64/perl5/vendor_perl /usr/share/perl5/vendor_perl /usr/lib64/perl5 /usr/share/perl5 .) at /tmp//mount.pl line 16. BEGIN failed--compilation aborted at /tmp//mount.pl line 16. -
Got it onto a CentOS 6 box, the script runs fine there (once you install Perl, who uses Perl anymore?)
-
It works, on the disparate CentOS 6.6 system I was able to do the file restores!!
-
What is needed here is for the Linux systems to be treated like the Windows ones and be given a way to mount them over a file protocol rather than a block one (both options is ideal.) For Windows they use SMB. This would be fine for Linux too, but not ideal Since the Unitrends device is Linux too, no reason to not use NFS which would be ideal for Linux. Or do NFS as well as SMB since SMB is already there. That would be the best. But NFS is what I think most Linux admins would want, nearly 95% of the time. It's native, it's fast, it's easy and chances are it is already there.
-
I ran into an issue not too long ago where my Unitrends didn't seem to be backing up everything with its masters. Ran a new master manually and that fixed it. I guess this is where good practice recoveries come into place.
-
@thanksajdotcom We were actually getting consistent failures across the board on all the VM's, seemingly due to an issue within VMware itself. However, I noticed we were using incremental forever on all of the VM's, and I was curious how the issue could be VMware needing an update. I ran manual fulls on the VM's, and all but one succeeded. So, I swapped out the incremental forever (awesome when it works, but honestly just has too much voodoo and pixie-dust for me to ever bother troubleshooting it) and set up a weekly full/nightly incremental strategy. So far, all but the one are working now. That's another issue, which I am currently addressing. As far as the incremental forever... it's "Unitrends' best practice", but sometimes environmental factors just don't play well with it, and it's a ton easier to go with a legacy backup strategy than worry about it, IMO.
-
@art_of_shred said:
@thanksajdotcom We were actually getting consistent failures across the board on all the VM's, seemingly due to an issue within VMware itself. However, I noticed we were using incremental forever on all of the VM's, and I was curious how the issue could be VMware needing an update. I ran manual fulls on the VM's, and all but one succeeded. So, I swapped out the incremental forever (awesome when it works, but honestly just has too much voodoo and pixie-dust for me to ever bother troubleshooting it) and set up a weekly full/nightly incremental strategy. So far, all but the one are working now. That's another issue, which I am currently addressing. As far as the incremental forever... it's "Unitrends' best practice", but sometimes environmental factors just don't play well with it, and it's a ton easier to go with a legacy backup strategy than worry about it, IMO.
Good to know. I have a client I was helping with that is using a Unitrends Appliance. Most all of the backups are incremental forever. And about once a month or they will say it's missing the master and not be able to backup. Unitrends has just told them to re-run the master when it does this but, of course that's becoming too much of a pain for them. I guess I'll just advise them to switch over to a normal full/incremental backup schedule. Sad that the marketed feature isn't really working.
-
It really does work most of the time. It's difficult to make a plug and play, one size fits all product for business computing. Every environment is unique, and every admin has their own methodology to configuring their architecture (often, that is driven by general cluelessness and/or ignorance). I think that, given the diversity of what you have to be able to adapt to, the roughly 95% success rate that I have seen with incremental forever is pretty decent. The thing that makes it somewhat aggravating is that it is a proprietary mechanism that is held rather tightly. Even inside of Unitrends, there doesn't seem to be a lot of general knowledge floating around about how to fix it when it doesn't work. The algorithms that control it are basically "unknown". It's a magical thing, powered by pixie dust, and you don't mess with it; it just kinda does its thing. If you have a Unitrends support contract, and it's giving you trouble, they can help diagnose it and get it fixed. For the rest of us...
-
@art_of_shred said:
@thanksajdotcom We were actually getting consistent failures across the board on all the VM's, seemingly due to an issue within VMware itself. However, I noticed we were using incremental forever on all of the VM's, and I was curious how the issue could be VMware needing an update. I ran manual fulls on the VM's, and all but one succeeded. So, I swapped out the incremental forever (awesome when it works, but honestly just has too much voodoo and pixie-dust for me to ever bother troubleshooting it) and set up a weekly full/nightly incremental strategy. So far, all but the one are working now. That's another issue, which I am currently addressing. As far as the incremental forever... it's "Unitrends' best practice", but sometimes environmental factors just don't play well with it, and it's a ton easier to go with a legacy backup strategy than worry about it, IMO.
@Jaguar said that best practice is incremental forever only on servers with over 1TB of data. Personally, I think if it's more than 500GB then it's justified. But the synthesizing takes a lot of resources, so doing full w/ incrementals avoids that on servers like straight DCs that are small.
-
@art_of_shred said:
It really does work most of the time. It's difficult to make a plug and play, one size fits all product for business computing. Every environment is unique, and every admin has their own methodology to configuring their architecture (often, that is driven by general cluelessness and/or ignorance). I think that, given the diversity of what you have to be able to adapt to, the roughly 95% success rate that I have seen with incremental forever is pretty decent. The thing that makes it somewhat aggravating is that it is a proprietary mechanism that is held rather tightly. Even inside of Unitrends, there doesn't seem to be a lot of general knowledge floating around about how to fix it when it doesn't work. The algorithms that control it are basically "unknown". It's a magical thing, powered by pixie dust, and you don't mess with it; it just kinda does its thing. If you have a Unitrends support contract, and it's giving you trouble, they can help diagnose it and get it fixed. For the rest of us...
Yeah they have a support contract but still haven't fixed it. They do call Unitrends and they re-run the masters for them. I've stayed out of it though since that's not what they pay me to do. I just deal with the switches/routers. So all of this is hear say.
-
@art_of_shred said:
It really does work most of the time. It's difficult to make a plug and play, one size fits all product for business computing. Every environment is unique, and every admin has their own methodology to configuring their architecture (often, that is driven by general cluelessness and/or ignorance). I think that, given the diversity of what you have to be able to adapt to, the roughly 95% success rate that I have seen with incremental forever is pretty decent. The thing that makes it somewhat aggravating is that it is a proprietary mechanism that is held rather tightly. Even inside of Unitrends, there doesn't seem to be a lot of general knowledge floating around about how to fix it when it doesn't work. The algorithms that control it are basically "unknown". It's a magical thing, powered by pixie dust, and you don't mess with it; it just kinda does its thing. If you have a Unitrends support contract, and it's giving you trouble, they can help diagnose it and get it fixed. For the rest of us...
Only complaint would be that for a backup system, 95% success rate is way, way too low for it to be a recommendation. It should be a "this is really fragile but if you want to give it a shot and monitor it closely, here it is" kind of thing at that point. It's a neat idea but if it doesn't match the reliability of traditional setups, I'd think recommendations should fall to the reliable. No aspect of IT has a stronger leaning towards consistent, conservative and reliable as backups.
