Incremental backup
- Thread starter al79
- Start date
I don't really know how to do this. I tried uploading an attachment, but the page says "Uploads not utilized within one hour will be deleted" which makes the whole concept of uploading files pointless. And I tried to PM you, but it says I do not have the rights to do that. So I gave up and registerd on a free file hosting site: box.net. Here is the link to the current alpha version:
http://www.box.net/shared/7c8rbfbxm8
I hope this work, but it's the first time I use this site, so I might have done something wrong...
This script neither does backups, nor restore. It takes existing .tar backups created by vzdump, and either creates a differential file from it, or gzip it, if "CYCLE_DAYS" is over. It can also recreate a full backup from the differential one.
My tips:
1) You need xdelta3 installed. "apt-get install xdelta3" did it for me...
2) Read the whole script before using it, so that you understand what it does.
3) Use "--debug --dry-run" first.
4) Try it on a test directory, and don't use huge backups to test.
5) It expects the standard vzdump file naming.
6) You can only recreate the full backup from a diff if you still have the "base" backup!
7) It won't work if you let vzdump delete files; I use "--maxfiles 1000" so that vzdump does not delete anything. And it is more efficient if vzdump does NOT compress itself, because you need to decompress to get the difference. My script does the compressing itself afterward. xdelta3 own auto-decompression is buggy, which is why Proxmox team stopped using it. But I got around that by doing the compression myself.
8) Bug reports and suggestions welcome!
http://www.box.net/shared/7c8rbfbxm8
I hope this work, but it's the first time I use this site, so I might have done something wrong...
This script neither does backups, nor restore. It takes existing .tar backups created by vzdump, and either creates a differential file from it, or gzip it, if "CYCLE_DAYS" is over. It can also recreate a full backup from the differential one.
My tips:
1) You need xdelta3 installed. "apt-get install xdelta3" did it for me...
2) Read the whole script before using it, so that you understand what it does.
3) Use "--debug --dry-run" first.
4) Try it on a test directory, and don't use huge backups to test.
5) It expects the standard vzdump file naming.
6) You can only recreate the full backup from a diff if you still have the "base" backup!
7) It won't work if you let vzdump delete files; I use "--maxfiles 1000" so that vzdump does not delete anything. And it is more efficient if vzdump does NOT compress itself, because you need to decompress to get the difference. My script does the compressing itself afterward. xdelta3 own auto-decompression is buggy, which is why Proxmox team stopped using it. But I got around that by doing the compression myself.
8) Bug reports and suggestions welcome!
file upload is enabled in the forum (PM is disabled). if you upload a file and you do not attach it to a post it will be deleted in one hour. but not if the file is linked to a post.
if it does not work like this its a bug and I will correct it - can you try it again?
OK, I tried again. From both the "normal" and the "basic" uploader I get the message:
Code:
The following errors occurred:
diffBackup.py: Invalid FileThen I tried to zip it and it finally worked; does the uploader only accepts archives?
Well, I'm not using those, but I guess that if the backup is one large file, my script could be modified to do the same with it. The critical point is: you can't diff a GZ file (or more precisely, you could, but the implementation of the binary diff program is bugged and it doesn't work). That is the main reason why the script is somewhat complicated.