We are aware of the issue with the badge emails resending to everyone, we apologise for the inconvenience - learn more here.
Forum Discussion
Backblaze is backing up Smart Sync’d files
I turned on Smart Sync (in our business account) and it seemed to be working as promised, until Backblaze kicked in and started backing up all 1.8TB of our “online only” files. What gives?
- When a file is marked as online only, if that file is accessed, Dropbox will download the file and mark it as Local. This is how you're able to open a file that's marked as online only.
Best guess, BackBlaze is trying to access the file to back it up, so Dropbox is downloading it and marking it as Local again. You'll probably need to add an exclusion in your backup set for those files/folders that are marked as online only.
Ok, that makes sense.
I’ll add an exclusion rule for the affected folders in the Backblaze client, but I should probably contact them about this scenario. If nothing else, it’s something that Blackblaze users should be made aware of.
Thanks for the speedy reply.
I'm having a more severe case of this problem.
Backblaze seems to THINK it backed up my SmartSync files, but when I try to restore them from Backblaze those files are corrupt. I've tested this on a variety of about 5 different files.
This is a bad situation because I don't simply have a SmartSync folder that I can ignore, but I selectively keep certain files offline and others online based on my needs. Dropbox also has a setting to keep files online or offline depending on how recently you edited it. So for instance in any given folder some might be online and others might be offline, and in your backups you would have no way of knowing which are there and which are not.
I've been in contact with BackBlaze but so far still at the stage of helping them realize there is not just a single folder I can tell it to ignore. And also trying to get them to acknowledge that having files in the backup which appear to be full size and valid but that in reality are corrupt is not a great situation.
Ideally, they would be able to detect that a file is offline vs online and then give it special handling. They could either ignore it, or back up a stub and mark it as such in their system, or try to download and backup the actual file and only if successful represent it as backed up.
Does anyone know whether backblaze detecting it is feasilbe? For instance is there a flag of some sort that Dropbox sets (ideal!) or could they notice that the actual size on disk is different than the represented size. This information is available to the user, as shown in my screen shot, so I hope it would be available to Backblaze.
(edit: I'm not allowed to upload an image but you can see for yourself if you select Get Info in the Finder for a file which is smart synced to be offline)
I work at Backblaze on the client that does the backups. Can you reach out to Backblaze support and get them some info on this issue:
> when I try to restore them from Backblaze those files are corrupt
I'm trying to figure out how that is possible, and what you mean by "corrupt"? Are you on Window or Macintosh (you said "Finder" so I'm guessing Macintosh), and by "corrupt" can you provide a correct file and a "corrupt" file for me/Backblaze/support to examine?
Backblaze is pretty profoundly simple - it looks for file modification times that have changed, and if they have changed then Backblaze reads the file into RAM, encrypts it, and sends it to Backblaze. I'm trying to figure out how Dropbox can function AT ALL if reading the files does not result in the correct original bytes. My first guess is your files are corrupt locally, in Dropbox, and in Backblaze - consistent.
> Does anyone know whether Backblaze detecting it is feasible?
We have a good relationship with Dropbox, and we can probably work that out between our engineers. The only question we have is: "what is the correct behavior customer wants?" If you want those files backed up, then they pretty much have to be downloaded to be local so Backblaze can read them. If you NEVER want anything in Dropbox to be backed up, hopefully if you exclude them in your Backblaze interface and they won't be backed up (or downloaded locally).
Hi @brianwski,
Thanks for your response! I’m replying from memory here to explain the issues, but if you’re serious about addressing this I’ll be happy to verify and provide details/testing etc.
With Dropbox’s SmartSync, certain files and folders will be kept in the cloud, but will appear to the user as if they are on your hard drive. The key feature here is that these cloud stored files are not simple in a folder, they are intermixed alongside regular files all throughout my Dropbox hierarchy. So I was having problems making this point to backblaze support because IIRC the solutions offered were all along the lines of telling me to “exclude that folder” but there is no specific folder unless I exclude the entire dropbox!
I’m glad that you have a good relationship with Dropbox because my dream solution (#3 below) would probably be easiest in coordination with them.
Solutions:
1. The default seems to be that each time BB runs across a smartsynced file, it tries to download the file and then back it up to backblaze. This is problematic for two reasons (a) because apparently glitches occur where it backs up the file that is not (yet?) fully downloaded, with no error message. When I attempt to restore the file I get some incomplete/corrupt version. (b) it’s also problematic because it causes all those offloaded files to be reloaded to my hard drive, defeating the space saving benefits of the smartsync
2. You could simply recognize smartsynced files and not back them up. This should be fairly straightforward since a user can determine that a file is smartsynced by the size showing in the get info window.
3. My dream solution would be that you recognize those smartsynced files, and then (with my permission) go get them directly from Dropbox rather than having to download and the reupload through my system. This would save me time and bandwidth, avoid defeating the purpose of the smartsync system, and presumably avoid the corruption issue that I described.
> 3. My dream solution would be that you recognize those smartsynced files, and then (with my permission) go get them directly from Dropbox
We were discussing this internally at Backblaze on Friday, and came up with the same proposal for an "ideal solution". Everybody agrees it would be a great solution. Let me see what we can do.
-- BrianW
