Dropbox Kept Files Around For Years Due To 'Delete' Bug (bleepingcomputer.com)

Dropbox has fixed a bug that caused old, deleted data to reappear on the site. The bug was reported by multiple support threads in the last three weeks and merged into one issue here. An anonymous Slashdot reader writes: In some of the complaints users reported seeing folders they deleted in 2009 reappear on their devices overnight. After seeing mysterious folders appear in their profile, some users thought they were hacked. Last week, a Dropbox employee provided an explanation to what happened, blaming the issue on an old bug that affected the metadata of soon-to-be-deleted folders. Instead of deleting the files, as users wanted and regardless of metadata issues, Dropbox choose to keep those files around for years, and eventually restored them due to a blunder. In its File retention Policy, Dropbox says it will keep files around a maximum 60 days after users deleted them.

Comments

Correct title

[Anonymous Coward]

Dropbox Kept Files Around For Years Due To Delete 'Bug'

FTFY

Re:Correct title

[Anonymous Coward]

I wonder if other cloud service providers have such 'bugs'.

Re:Correct title

[AmiMoJo]

Probably. Most cloud services rely on duplication to secure user data. The better ones geographically distribute the copies in case the datacentre goes up in flames.

Then you have Nearline storage, and offline storage copies.

When a file is deleted, all the copies have to be removed. Bugs, failed and inadequately wiped failed/recycled disks, cached copies... It can easily go wrong.

Which is why you don't store anything sensitive unencrypted in the cloud.

Re:Correct title

[m2shariy]

This is a feature which allows to undelete data during grace period. Every provider must have that. This specific issue is that 1. the grace period was way too long. 2. the data was accidentally undeleted.

Re: Correct title

[Anonymous Coward]

A bug let users know their files haven't been deleted. The bug didn't keep the files, that was a decision by someone in a suit.

Re:Correct title

[elrous0]

I wonder if other cloud service providers have such 'bugs'.

Only the ones on their knees sucking NSA and FBI dick. In other words, pretty much all of them.

But kudos to Dropbox for their incompetent slip that confirmed everything we'd feared about such file sharing services archiving and sharing data with the government.

I wonder if they're doing this for the Chinese government too. I suspect the answer is a resounding "yes."

Re: Correct title

[Anonymous Coward]

The bug didn't keep the files, that was a decision by someone in a suit.

No doubt a suit with a FBI or NSA security badge attached to it.

Re:Correct title

[Anonymous Coward]

I find it hard to believe that nobody noticed 8 fucking years worth of "deleted" data until now. This was done on purpose and noe Dropbox is trying to use the scapegoat of "it's a bug".

Yeah, bullshit.

Re:Correct title

[Immerman]

Was it 8 years worth of data, or just data from 8 years ago? Big difference, and the summary at least indicates the latter.

Re:Correct title

[AHuxley]

If it was the FBI nobody would be allowed to talk about the files left in place as part of an ongoing investigation.
Every ip requesting that file would be logged over the years.

Re:Correct title

[stooo]

>> Dropbox has fixed a bug that caused old, deleted data to reappear on the site.

Now, your deleted data only appears to the NSA, where it belongs.

Re:Correct title

[JustAnotherOldGuy]

Was it 8 years worth of data, or just data from 8 years ago? Big difference, and the summary at least indicates the latter.

I think it's awfully trusting of you to believe that they weren't saving everything deliberately.

It seems ridiculously unlikely that this sort of thing could go on for 8 years without anyone noticing. If nothing else, the need for constantly increasing storage should have made someone wonder what was going on.

But seriously, one of their key functions, namely "deleting a file" didn't work and no one noticed for almost a decade?

Re:Correct title

[The-Ixian]

It seems ridiculously unlikely that this sort of thing could go on for 8 years without anyone noticing. If nothing else, the need for constantly increasing storage should have made someone wonder what was going on./quote

Yes, exactly what I was thinking. The cat's out of the bag now. Dropbox can say whatever they want, the evidence is pretty clear: files were kept for 8 years at least (otherwise they couldn't be restored).

I am sure there are lots of reasons for cloud providers not to delete your data. It is just too valuable a tool to delete. You can gain all kinds of insights about your users from the data they upload. Whether their motivation is benign or not, it doesn't matter, the truth is that once a file leaves your computer it is no longer yours.

Re:Correct title

[Cro Magnon]

I don't have a problem with the 30-60 day grace period that Dropbox claims to have. I do think 6-8 years is too long! :p

Re:Correct title

[GuB-42]

I think we can use Hanlon's razor here "do not attribute to malice what can be explained by stupidity".
It is not unthinkable at all that undeleted files go unnoticed. It happens all the time where I work. And because storage needs constantly increase anyways, it isn't that noticeable.
For example, files may be copied to A, but because of some problem, the get copied to B instead. Later, the problem is fixed and files are copied again to A, where they should be. B is left untouched. Now the users wants to delete its files and they also get deleted from A, as expected, but there are still some left on B, and they aren't deleted because they shouldn't be there in the first place. Maybe later, perhaps after a disk crash, some restoration script goes back to B and starts restoring the files, causing the deleted files to resurface.
Sysadmins really want to make sure they don't lose data, as they should. So if they have to chose between risking losing data and improperly deleting files, they will most likely chose the second option.

Re:Correct title

[Immerman]

Exactly what would cause you to casually notice that the random jellybeans that *someone else* meant to take out of a jar were still there?

If it were a systematic oversight, and jellybeans were *never* (or rarely) actually removed, then the size of the jar would need to grow continuously as new beans were added and the old ones weren't removed. Of course the jar would need to steadily grow regardless, because every year people are adding more and larger beans than the year before, but the difference in the rate of growth would likely catch someones attention.

If it were a one-time thing though - or just a rare occurrence, say 10% of the time the jellybean removal chute malfunctioned and dumped the beans back in the jar, then there would be no really obvious sign that anything was amiss.

Re:Correct title

[JustAnotherOldGuy]

It is not unthinkable at all that undeleted files go unnoticed. It happens all the time where I work.

Really, you guys lose track of petabytes of data and no one notices for the better part of a decade?

Please tell me who you work for so I can avoid them like the plague.

Re:Correct title

[JustAnotherOldGuy]

Exactly what would cause you to casually notice that the random jellybeans that *someone else* meant to take out of a jar were still there?

Oh gosh, I don't know...log files? Sanity checks? System audits? Function testing? The fact that the jellybean jar is now the size of my local Wal-Mart?

Seriously, if you don't see why this whole "oops-we-fergot-to-actually-delete-yer-file" thing is super suspicious, then there's probably nothing I can say that would make it clear.

Re:Correct title

[Immerman]

Suspicious? Absolutely.

I just think it's worth pointing out that when you're dealing with as much data as they do, they would have to be extremely thorough to notice that something that was supposed to be deleted is still hanging around - I mean it's not the sort of thing that any users will send in a bug report on.

Re:Correct title

[GuB-42]

I won't tell you who I'm working for but it is not an IT company, which is a good thing since the IT department is abysmal...
However, I think you overestimate companies in general. You know, you, as an individual, forget about things sometimes. Have you ever found something in house house you thought was lost? Have you ever forgotten to pay a bill? Have you ever missed a deadline just because you forgot about it? If you are a normal human, you probably ended up is such situations.
Companies are made of people, they make the same kinds of mistakes. It is even worse because people tend to join and leave the company, information is lost between as it travels the management chain, etc... Proper methods help alleviate the problem but it is far from easy. I've seen many, many facepalm moments, from companies that should know better, both first hand and in the news. And these are companies that are still alive and profitable.

So I think it is very possible for petabytes to go unnoticed, though I think a few terabytes are all what it takes for the bug to appear. And years won't save you, the older things are, the more likely you are to forget.

Re:Correct title

[JustAnotherOldGuy]

Have you ever found something in house house you thought was lost?

Yes, but I've not "found something" I thought was lost 600 times.

-

Have you ever forgotten to pay a bill?

I have, but I haven't forgotten to pay 600 million bills.

-

Have you ever missed a deadline just because you forgot about it?

Sure, but I've never missed 600 million deadlines.

-

DropBox

[Anonymous Coward]

So then, DropBox will not Drop. The Box?

Whoopsie!

[Anonymous Coward]

Anyone else surprised that an online mega-corp has lied about it's data retention policy?

Yeah sure.

[Anonymous Coward]

They didn't notice terabytes of data just piling up over 8 years. Mkay.

Re:Yeah sure.

[Anonymous Coward]

Probably closer to multiple petabytes. So yes, they knew about this and made a mistake that showed their lies.

Oops

[Anonymous Coward]

Sorry, it was uncovered that we don't delete anything ever and someone found a bug to find those files you thought were deleted. We still don't delete in case the feds want to take a peak at anything, but we've fixed the bug so they won't appear to you anymore. That way you'll think your files are deleted.

Re:Oops

[KiloByte]

We still don't delete in case the feds want to take a peak at anything

As keeping deleted files comes at a significant cost, my guess is that malice here doesn't come from Dropbox itself.

Re:Oops

[Anonymous Coward]

probably from their funding provider, the US taxes supported agencies.

Re:Oops

[Anonymous Coward]

This went "unnoticed" for years. Not likely. That would be like you not noticing the Olympic size pool that your neighbor dug in your back yard, after he re-routed his driveway, over your property so that his yard was larger as well as connecting his electrical box to your connection and has been getting free electricity for years.

Re: Oops

[Anonymous Coward]

I moved into a rental house once. After a couple month of $20 electric bills summer came up. The AC wouldn't work. The problem? The previous renter pulled off the meter and jumped the connection bypassing the meter on one of the connections. The AC was unable to get the current it needed thru that hack job.
Things go unnoticed. Well, at least until someone who cares shows up. Don't count on that. Be the one who cares. That stupid hack could have burned the house down and killed somebody.

Re:Oops

[Excelcia]

Storage is cheap. If you think that Dropbox ever deletes anything you store there, then you are naive.

In general, if anything is free, then you are in some way the product. If data storage is free, then your data is the product. I highly recommend Syncthing. People need to keep ownership of their data.

Yet Another

[Anonymous Coward]

Yet another cloud service provide lying about the service it provides. Whatever happened to truth in advertising laws?

Re:Yet Another

[elrous0]

Whatever happened to truth in advertising laws?

They got thrown out with the Patriot Act and a million other laws which have turned our government into Big Brother.

FTFY

[Anonymous Coward]

Instead of deleting the files, as users wanted and regardless of metadata issues, Dropbox choose to secretly keep those files around for years, but accidentally made this visible to the user when they restored them due to a blunder.

Tell me the OS

[unixisc]

Which OS is it that is so complicated that when you ask it to delete a file, it doesn't? I wasn't aware that one even exists

Re: Tell me the OS

[Zero__Kelvin]

I can't name an Operating System that *doesn't* leave the file when you delete it. Typically the file system pointer is deleted but the file itself remains untouched by the file delete code. The difference here is that there is no guarantee future disk write won't overwrite all or part of your file and retrieving it is quite a bit more tricky ;-)

Re:Tell me the OS

[93 Escort Wagon]

Which OS is it that is so complicated that when you ask it to delete a file, it doesn't? I wasn't aware that one even exists

I strongly suspect this has nothing to do with the OS and everything to do with Dropbox Inc.

I don't imagine they actually delete anything - they probably just set a "do not show to user" flag. It's probably still there, ad infinitum, along with any and all metadata connecting the file to you as an individual.

Re:Tell me the OS

[guruevi]

Pretty much any filesystem since before MSDOS only unlinks the file, not really deletes it. Windows 95 came with a Trash can feature that only moved files to the Trash until the user unlinked the files. These days cloud/flash based storage will do pretty much the same, keep the data around until it's either overwritten due to space congestion or deleted by an admin.

Re:Tell me the OS

[Dogtanian]

Which OS is it that is so complicated that when you ask it to delete a file, it doesn't? I wasn't aware that one even exists

As others pointed out, no, typical OSs don't overwrite it when you "delete" it.

In addition to this, however- I don't know what Dropbox's setup is, and I know sod all about enterprise storage et al. However, I feel pretty confident in assuming it's *not* going to be anything as simple as an "off the shelf" hard drive or even RAID setup using the standard Windows, Linux or whatever facilities and filesystems like one would find in a desktop PC!

The comparison is therefore pretty meaningless.

Re: Tell me the OS

[Anonymous Coward]

try S3

Re:Tell me the OS

[The-Ixian]

Pretty much any filesystem since before MSDOS only unlinks the file, not really deletes it.

Yeah, nice try.

Except that if this was the case, you would almost certainly not be able to restore 8 year old files. How many times do you think they have updated their storage systems in the last 8 years? I am guessing the answer is more than 0.

Re: Tell me the OS

[Dogtanian]

try S3

You mean Amazon S3? Not sure what your point is. That's not a local (or locally-controlled) storage option and Amazon aren't even open about how it's implemented.

Re:Tell me the OS

[guruevi]

I highly doubt they up and replaced the entire system migrating all live data to new systems. The problem with "cloud" file systems is that it's not really a file system, its a database (object store whatever you call it), deleting an "object" does nothing really but leave a hole in the database and such fragmentation cannot easily be resolved unless you devise a fitting algorithm. So most systems never delete anything truly, just mark it deleted and if ever they need space, they could "vacuum" the thing but it's such a pain in the neck or even a risk that it's never done. Even if you replace the underlying disk, the entire database gets replicated onto it including the "dead" space, neither the upper layers have any idea about the media nor does the media have any idea about the upper layers. Most of these "file" systems are very wasteful, fragile moronic thing cobbled together for a 2 year IPO goal not designed by seasoned programmers for the long haul.

Re:Tell me the OS

[The-Ixian]

Well, there you go. You made my point better than I (obviously) did.

This can't be a file system related thing.

Alternative Facts

[Anonymous Coward]

Funny how for exactly 8 years, this internet company managed to accidentally not delete documents that its users asked to be deleted in confidence, and a week after a new administration takes power, they magically find out that they weren't deleting any documents and now they have to be purged.

It's almost like someone wanted to keep these deleted documents around so they could comb through them to find patterns, or something. It's a good thing that our government isn't spying on us through our social media sharing sites, or something.

Re: Alternative Facts

[Anonymous Coward]

I read it the other way -- that OP is saying Obama's administration had been doing nefarious things but now there's a new sheriff in town who won't take kindly to that so they spilled the beans and tried to figure some way to fix things without taking blame or looking shady.

I don't think it's related to the government at all. It sounds like a really poor business decision to tell users the files are always deleted after 60 days, but then keep the data unreferenced to the public forever. You think someone on the board or somewhere would say "hey, that's not our data, we can't just keep it. there are legal liabilities to consider and no positive merit to keeping the data." guess not.

Re:Alternative Facts

[Anonymous Coward]

Yes. What you just said is a giant load of alternative facts.

Re:Alternative Facts

[spikenerd]

...whereas now, they are totally gone. I'm certain they didn't just "fix this bug" by hiding the data better.

Re:Alternative Facts

[antdude]

I always assume the hosts have copies even if you nuke them. :(

Re:Alternative Facts

[Anonymous Coward]

Same. Assume everything you share, even privately, is public.

Re:Alternative Facts

[Anonymous Coward]

Don't worry though. All of the online password storage sites people use wouldn't dream of doing anything shady.

Re:Alternative Facts

[The-Ixian]

Are you using your Jump to Conclusions mat again? I thought we burned that thing...

OK, new plan!

[Anonymous Coward]

1.) Each day, upload a bunch of random garbage data to max out available storage
2.) Delete aforementioned data
3.) Rinse
4.) Repeat

Eventually, the 'deleted' data will be incredibly burdensome to maintain.

Get a life

[Anonymous Coward]

Man a lot of butthurt "free" lameoid users on that support thread

Hooray!

[Anonymous Coward]

Now all those broken links on pirate video sites may start working again!

Delete 'Bug', not 'Delete' Bug

[citizenr]

Remember this - NOTHING gets deleted from the cloud, its just too precious.

Storage costs?

[Anonymous Coward]

You would think that Dropbox would mark data as deleted and let the storage space be recycled after some delay (to let users "undelete" files due to user error).

Given the scale of Dropbox and the amount of storage they need to buy, this level of "bug" is a deliberate design choice.

Re:Storage costs?

[guruevi]

Not necessarily, all of Dropbox is approx. 120-200PB. Distributed over thousands of storage servers it's really peanuts to save people's history (which are mostly small delta's). Running a storage system of 200TB myself, people tend not to delete stuff all that much and even so, the entire amount of people's previous storages is encapsulated every time we have to upgrade (every 3 years). 10 years ago we stored close to 10TB, now 10TB is a rounding error on the upgrade.

Probably kept on purpose

[TheOuterLinux]

It wouldn't surprise me if the files or metadata were kept on purpose. They'll be the next Yahoo for it though. God only knows what just info could be found. I'm sure intelligence agencies love it.

They can keep files forever

[Anonymous Coward]

I just don't want to see them if deleted. What Dropbox wants to internally do with it is up to them. If you're putting sensitive files on a public cloud device, encrypt them.

I setup Dropbox primarily for sharing files among our group. With the Dropbox Windows interface when some people drag files from a Dropbox directory to a 'local' directory they don't realize it's a move from Dropbox (which Windows considers as local directory) rather than a copy (if Windows thought it was a network directory). When I noticed a large body of files went missing from Dropbox I asked whoever moved them to put them the fuck back. Nobody fessed up, so I blamed Dropbox. They not only restored the files, they told me who (machine name) moved what files and when. I got my free storage money's worth.

example #56,743,465

[Anonymous Coward]

why not to use 'the cloud' for ANYTHING

another reason..

[e432776]

..to keep your data on your own machines. In that case the "delete" bug is the more usual variety, where files are accidentally lost or clobbered. These cloud providers are all upside down!

It was a real bug, now fixed.

[misnohmer]

It sounds like the bug was that the files re-appeared. That is now fixed, so most likely we're back to only the NSA and some generous customers having access to those "deleted" files.

Correct Issue

[fyngyrz]

No, the specific issue is that if you put your data in "the cloud": It's out of your control. You've put your trust in people who are out of your control and who are almost certainly motivated entirely by money and power, not your well-being or security, except as that drives the first. Data storage providers can -- and will -- do things with your data without telling you that are completely out of your control. Including hand over the data to any entity that can apply enough monetary or threat pressure to motivate them -- like a government or an advertiser.

Unless the data is of absolutely no consequence, putting data in "the cloud" is a very poor decision.

"The cloud" is a touchy-feely name for a monumentally risky choice in data storage. More honestly, it could be called "Untrustworthy storage." Even that's a little too friendly.

Re:Correct Issue

[The-Ixian]

You are right, of course. However, the same could be said about bank safety deposit boxes as well, but those are generally accepted as "safer than your mattress" (which is entirely under your control) even though law enforcement can get access to them.

A matter of point of view

[volodymyrbiryuk]

A bug to the customers, a feature to the government.

Wait till they hear about Alexa and Cortana and ..

[140Mandak262Jamuna]

You think the "voice assistants" Siri, Google Assistant, Alexa, Cortana etc who continually listen to the microphone do not save what they hear? You think the companies are not saving all that audio? Recently there was an article about ultra low bit rate audio codecs, tuned to human speech, that can record 80 years of audio in a 8 GB file.

Wrong

[JustAnotherOldGuy]

"Dropbox says it will keep files around a maximum 60 days after users deleted them."

Obviously this is wrong, and to suggest that Dropbox had no idea that this was happening seems a bit naive, no?

Deleting files is one of the primary bits of functionality that Dropbox has; to think that somehow they flubbed the code to remove a file is, to me, flatly unbelievable.

Called it.

[Anonymous Coward]

I thought something like this would happen as soon as I saw that Condoleeza Rice was on DropBox's board of directors.

People still use Dropbox?

[x_t0ken_407]

Lol! If you're still using Dropbox after Snowden's revelations (I never used them, FWIW), you deserve this. Seriously, network your own box and install Owncloud, or whatever other "cloud" shit interface you need -- if simply rsync'ing files to a headless box is to technical for whoever needs the data.

Disclaimer: I don't know all the use cases of Dropbox and I know some are forced to use it...you guys are cool, I guess.