German Health Insurance Card CA Loses Secret Key
Christiane writes "The SSL Root CA responsible for issuing the German digital health insurance card lost its secret private key during a test enrollment. After their Hardware Security Module (HSM) dutifully deleted its crypto keys during a power outage, it was all 'Oops, why is there no backup?' All issued cards must be replaced: 'Gematik spokesman Daniel Poeschkens poured scorn on the statement that Gematik had insisted on the service provider carrying out a test without backing up the root CA private keys. "We did not decide against a back-up service. The fact of the matter is that the service provider took over the running of the test system, so it also has to warrant its continuous operation. How it fulfills this obligation is its own responsibility."'"
Not even a month ago you chided them because there were too many copies (some of them even offsite, they just didn't know who had them now), now you chew them out for having too few. Make up your effing mind!
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Even when accessing key material? C'mon! The Confidentiality, INTEGRITY, ASSURANCE triangle seems to be missing a couple of legs, in this instance.
That's really amateur. Sounds like someone swapped the Smart Cards with Dumb Cards...
"Speaking the Truth in times of universal deceit is a revolutionary act." -- George Orwell
Once again, misleading title to a different summary.
For fuck's sake, the Germans didn't lose the key.
The SSL Root CA lost that.
Get the facts right.
For a second i was wondering how Germans could that stupid. That is unlike the Germany i know. And exactly as i suspected, the German insurer had been insisting the root CA for backup while the CA thought it was unnecessary.
Is it the German company's fault?
"Doing what i can, with what i have." ~ Burt Gummer
Maybe they should check with the NSA or CIA? They've got a backdoor into EVERY system, and may still have the key saved on a laptop lying around somewhere.
I'd rather the key be lost, than stolen, hacked, made-public, etc. At least it didn't breach security in the typical manner.
Is the cost of re-establishing the chain of trust (ie a new root and replacing all of the cards) higher than the value of the data that this system was protecting?
It would easily be found be searching the nearest pub car park for USB keys, or checking the train that the relevant civil servant travelled home on.
The entire concept of PKI revolves around the inheritance of trust from the root CA. It seems pretty clear these guys can not be trusted. I would be worried about the people who have to use them.
There are two fundamental ways to fail as a CA. There must be exactly one party in effective possession of the private key of the root cert. If the number of parties becomes less than or more than one, fail.
Mistakes happen, of course, and certificate infrastructures can be enormously complex. But if you're going to do any kind of risk mitigation, the absolutely most basic place to start would be with these two scenarios.
Parity: What to do when the weekend comes.
Q: How do you learn every German swear word in about 20 seconds?
A: Tell the German admin that you lost the root key.
stuff |
What I do for a living: Build a GPS mobile game
I'm confused, isn't this sort of problem exactly why you carry out system tests?
Sending out new cards to card testers during a systems test is hardly extraordinary.
The next time you have a thought, just let it go.
(1) outsourced government works even less well.
(2) exceptions are covered under rule #1.
if this is supposed to be a new economy, how come they still want my old fashioned money?
Undoing moderation in error....
"the firm's managing director, told heise online that following a voltage drop, something happened in D-Trust's "Trustcenter" that does occasionally occur" You cannot even say what's worse: A voltage drop even reaching the HSM or the HSM going suicidal and loosing the key. And all of that "occasionally"? Everytime they make popcorn in the microwave? As a german I am quite flabbergasted by this lack of german engineering, in one of the countries largest trust-centers.
Poeschkens claimed, "I know nothing! noth-thing!" and proceeded to blame the problem on a man he would only identify as "Hogan."
Bark less. Wag more.
Once again, misleading title to a different summary.
How's that again?
For fuck's sake, the Germans didn't lose the key.
The SSL Root CA lost that.
Hmm.. I wonder if that's why the title reads "German Health Insurance Card CA Loses Secret Key"?
What are you ranting about? The title says exactly what you say it should say, and then rant about how it's wrong...
Reading comprehension - you fail it!
Although I must laugh about how your post demonstrates exactly what you are claiming is wrong with the article's summary and title.
For those of you who are wondering what CA is, it stands for Certificate Authority. You see, the Germans have a hard time functioning without a constant stream of praise, so they have this authority in place that prints and sends certificates to people. Every day thousands of Germans get congratualted for crossing the street, for finding their car keys or for eating their 1000th potato of the month. You know you've walked into a German household when you see the wallpaper of framed certificates.
The problem here is that the company deleted the certificate-printing program since they thought someone was trying to hack in and print more certificates for themselves- no one is THAT special so they had to stop him. They forgot to have another program ready to print more certificates, so now Germany is under threat of entering a depression since they no longer get certificates telling them how special they are.
On a serious note: I don't follow this article very well with all the acronyms being spelled out but not explained, and no background knowledge of anything going on here. If someone would care to explain what is going on here to someone that has never heard the term CA, you should get a +5 informative easily.
My webcomic
And please don't forget to flush.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Best practices about CA management says you should have your secret key in a (physical) safe. Better yet, divide it in two pieces and put it along the passphrase in three different safes (part1+pass,part2+pass,part1+part2), so you can't lose key access even if you lose one safe, and nobody can take the key by opening a single safe.
Got Pike?
They can have my copy, I hacked their servers last week ;)
MP3 Search Engine
When they kick you out of the pub it's time to be at the curry house.
In talking with people (or company representatives) about their security regarding passwords and keys, I always told them two things.
First, all security experts will tell you that you should not keep copies of that stuff around.
Second, that's not a realistic expectation, stuff happens. The IT guy goes on vacation, has an accident, or dies. (Seen all 3 numerous times.) You fire the Admin for some reason. This building burns down. Etc.
A reasonable thing to do, is keep a password/key log with that critical information that is kept up to date at all times. You have two copies of it. Both are kept secure in good quality safes (not a $200 lockbox).
Both safes are in different physical locations, at least separate buildings, preferably miles apart.
The reason for this is pretty easy. Once again, things happen. I've seen buildings burnt down, flooded, inaccessible due to chemical hazards from a truck wreck, etc. You don't know what will happen, but if you have them stored at separate physical locations, you at least know you will be able to get to one of them if you need to, assuming nobody uses a nuke.
It all falls under that old techie saying, "So, when did your data become important to you? Before or after you lost it...".
Perhaps they could save their private key in the same place KFC and Coke do for their secret recipes. These guys know security!
I literally read that as scoured porn...
"We did not decide against a back-up service. The fact of the matter is that the service provider took over the running of the test system, so it also has to warrant its continuous operation. How it fulfills this obligation is its own responsibility."' If this were originally in English, it would mean "We knew this would happen and we tried to tell them, but those arrogant SOBs thought they knew it all and didn't want to listen to us. So we shut up, pulled up a chair, got some popcorn, and waited for the fireworks". I'm not sure that translates, though...
All issued cards must be replaced
...why? Unless they (along with everyone else) have lost their public key, there should be no problem verifying all previously signed cards.
1000th potato of the month? That's at least 33 potatos a day! That'd be a world record I suspect and well deserving of a certificate!
Read what I mean, not what I wrote.
Making multiple backup copies protects against losing the secret (the root key in this case), but clearly increases the risk of theft. Secret sharing is the way to backup and still be secure. In a "k out of n" secret sharing system, the secret is divided into n pieces, any k of which allow perfect reconstruction of the secret. What's amazing is that any k-1 tell absolutely nothing about the secret! The easiest to understand is a k-1 out of k system. For example, taking k=5 and assuming the secret is 1000 bits long, the first four pieces of the secret are totally random bit strings, each 1000 bits long. The fifth piece is the XOR of the secret and these four strings. It's not hard to see that any four pieces tell nothing, but all five produce the secret when XORed together. More complex k out of n systems are not too much harder to understand. For example, a 3 out of 5 system can be based on the coefficients (A,B,C) of a quadratic function y = Ax^2 + Bx + C. The coefficients can be determined by any three points (x,y) which lie on the graph. If C is the secret, and the 5 pieces of the secret are five points (x1,y1), (x2,y2), (x3,y3), (x4,y4) and (x5,y5) on the graph, then any 3 of them determine (A,B,C) and hence the secret C. But any 2 or less of them tell us absolutely nothing about C. Arithmetic is done in a finite field so that C is a bit string or similar. Martin Hellman http://www-ee.stanford.edu/~hellman/ http://nuclearrisk.org/
How it fulfills this obligation is its own responsibility.
bzzzt. wrong.
If you're the guy in charge, it's your duty to make sure things work. You can leave specifics to the contrator if you are sure, but as the saying goes, you can not delegate or outsource responsibility.
Assorted stuff I do sometimes: Lemuria.org
Actually it would be two words, because to properly curse a person we'd use an adjective in conjunction with a noun.
Mod parent up, please!
THe voltage-drop and the deletion of the key were both part of the tests they did. The HSM deletes the key because it thinks that the drop in voltage is a precursor to a physical attack (basically it thinks that it is being carried out of the datacenter). Everything worked as expected; they simply did not remember to make a backup...
Wait, THAT'S what happened? All this time I was willing to blame the incompetence of Gematik for not backing up the key and the vender just running some normal computer like tests. They still should of done this, but if the vender was purposely trying to delete the key then why the heck didn't they have a backup? This is just getting sad now. Next think you know, the next test they do, they will do a plain text export of the damn key for "backup"