|
zhar posted:I'm finally changing password manager from 1password to keepass (free!) and looking up good password practices it seems my master password isn't as secure as it should be. Or maybe it is as password databases have extra security to prevent brute force attacks, I don't know. What is the minimum strength the master password should be? I might just reuse my old one or make a new one but I don't want to overdo it as I'll be typing it a lot. Obviously I don't want to underdo it either. Is 4 random words enough? Automatic password security checkers basically just tick boxes, usually upper case, lower case, number, symbol, 8/12/16 characters. The more boxes your password checks, the better the score. This is not the same at actual security, which is about being resistant to guessing, which universally can be achieved through length alone (I'd actually argue that 66...66 with like 666 total sixes would be pretty drat secure, even if it's literally all the same number). So yeah, four random words is probably good, provides they aren't "I, am, a, cat". But also, dictionary attacks are a thing, so misspellings or using different languages isn't a bad idea.
|
# ? Mar 13, 2022 17:43 |
|
|
# ? May 30, 2024 18:04 |
|
You can reject a word or regenerate a whole phrase just fine. You want it random in that it’s not a meaningful sequence where the words are related, like “four-score-and-twenty” or whatever. You’re not going to break the magic security spell by swapping “pudendum” out for “daisy” or something.
|
# ? Mar 13, 2022 17:44 |
|
zhar posted:I will use the dice for 4 words but it feels uncomfortable having a master password significantly weaker than the passwords contained within, so I might add a fifth nonrandom word like my favorite word to make sure an attacker would be tripped up. Just lol if you don't use a verse from an 80s hair metal rock opera as your master pass phrase. Don't ya think that you need somebody? Don't ya think that you need someone? Everybody needs somebody You're not the only one You're not the only one Verify me
|
# ? Mar 13, 2022 17:46 |
|
Maybe try a generator without obscenity in it? Am gonna say thats one thing I miss about the old stupid systems where the sysop could see your password: such comedy reading what people used
|
# ? Mar 13, 2022 17:57 |
|
RFC2324 posted:The only properly secured password is to open to a random page of War and Peace, and use the entire page as your password Maybe the old school video game code wheels can make a comeback? We all know technology is cyclical. https://www.youtube.com/watch?v=S_Tz0YpDa6o
|
# ? Mar 13, 2022 18:07 |
|
Just write down 16 random ASCII characters and keep it in your wallet until you remember it (or gently caress it, forever - nobody who finds it is going to know what is for unless you get nabbed by a 3 letter agency).
|
# ? Mar 13, 2022 18:17 |
|
zhar posted:I'm finally changing password manager from 1password to keepass (free!) and looking up good password practices it seems my master password isn't as secure as it should be. Or maybe it is as password databases have extra security to prevent brute force attacks, I don't know. The database has far more protection against brute force attacks than normal websites and services. Keepass by default does 60000 loops of AES for key transformation. (ie your master pass isn't the "real" key to decrypt the database, whatever comes out the other side of that loop is.) That means 1 brute-force guess takes somewhere around 100,000x longer than against a website database leak that used SHA-256 password hashing. But you can go into the database settings on Keepass & set it to be much more difficult. There's a handy "1 second delay" button to make it take a full second of processing time, on whatever machine you are using. And you can also switch from AES to Argon2, which is resistant to GPUs because it needs memory. GPUs are only hyper-fast on stuff that fits inside their (tiny) cache. I'd recommend doing this! So that's how this: zhar posted:it feels uncomfortable having a master password significantly weaker than the passwords contained within (Don't actually do that though. Hardware improves all the time, so today's 30 seconds has a shelf-life. And also if you use keepass on a cell phone that 30 second delay becomes 2 minutes.)
|
# ? Mar 13, 2022 18:23 |
|
Klyith posted:explanation Thanks this is what I wanted to hear, I feel a lot more comfortable using the 'weak' 4 word password I ended up generating with nice benign words now. A bit too benign for me actually, I'm copying over the 1password wordlist to keepassxc as it's longer and generally has more interesting words than the diceware list which I think keepassxc uses. KillHour posted:Just write down 16 random ASCII characters and keep it in your wallet until you remember it (or gently caress it, forever - nobody who finds it is going to know what is for unless you get nabbed by a 3 letter agency). I like words because I can turn them into a nice allegory I can contemplate whenever I enter my password
|
# ? Mar 13, 2022 18:30 |
|
I’ve always found the best thing is to use something easy to remember but he’s so not you or your interest that no one in a millionaires could think of it. That said I prefer just using 1Password. I got introduced to it a job I used to work at and I just started using it for my home stuff and I just love it. I just wish it was a little easier to create new passwords and iOS sometimes. Plus since it’s used also for my own side work it’s a tax deduction.
|
# ? Mar 13, 2022 18:33 |
|
I had to set a password for a client's SSO and I don't want my personal PW manager on that machine so I generated a "more pronounceable" 32 character password but with added special characters, and wrote it on a piece of tape on the monitor. They also use 2FA and my IP is whitelisted, so I think I'm fine.
|
# ? Mar 13, 2022 19:13 |
|
Klyith posted:The database has far more protection against brute force attacks than normal websites and services. Keepass by default does 60000 loops of AES for key transformation. (ie your master pass isn't the "real" key to decrypt the database, whatever comes out the other side of that loop is.) That means 1 brute-force guess takes somewhere around 100,000x longer than against a website database leak that used SHA-256 password hashing. This is all true, but keep in mind that upping the difficulty of the key derivation like this is going to run into diminishing returns pretty quickly. To use your example, let's say you increase the number of rounds so that every attempt takes thirty seconds instead of one. Great, brute-force attacks are now thirty times harder! ...but you're going to have to twiddle your thumbs for 30 seconds every single time you unlock it yourself. (If that's something you do once a week, that wait might be no big deal. If it's something you do twice an hour, you're going to want to punch your screen before long.) A much more convenient way to add difficulty is to simply use a longer password. Even if you limit yourself to letters and numbers, each extra random character you add to the password gives you something on the order of six more bits of entropy, which translates to an increase in brute-force difficulty of 64x -- more than twice as good as the big annoying 30-second wait.
|
# ? Mar 13, 2022 19:56 |
|
i like to make them stupid trolls, either of the employer i'm resenting working at or the imaginary intelligence officer torturing you for it... "please water board me again" WHATS YOUR PASSWORD? PLEASE WATER BOARD ME AGAIN
|
# ? Mar 13, 2022 20:07 |
|
I have a few good randomly generated passwords that I can remember by the typing patterns, like for instance the general pattern is diagonally up and left, then down, then straight across. Jumps and upper/lowercase and symbols are of course scattered. Or as an alternative, use a pattern that you come up with, similar to //12donkey!!BREATH69\\. Sure, maybe it's easier to guess if someone knows the pattern you used, but there are countless patterns. You know the pattern, making it easier to remember. KozmoNaut fucked around with this message at 20:21 on Mar 13, 2022 |
# ? Mar 13, 2022 20:19 |
|
KozmoNaut posted:I have a few good randomly generated passwords that I can remember by the typing patterns, like for instance the general pattern is diagonally up and left, then down, then straight across. Jumps and upper/lowercase and symbols are of course scattered. Except for the fact that the pattern makes the problem size smaller 4-5 words randomly generated is way more memorable with a way bigger problem size IMO
|
# ? Mar 13, 2022 20:54 |
|
Buff Hardback posted:Except for the fact that the pattern makes the problem size smaller Only if the attacker knows the pattern. An attacker would have to know that a pattern was used and the format of the pattern. If not, they have to guess randomly and your password is like 24+ characters. Obviously use a secret pattern, and not necessarily one that's sort of symmetrical. Think of it as adding random characters between the words in a passphrase. KozmoNaut fucked around with this message at 21:20 on Mar 13, 2022 |
# ? Mar 13, 2022 21:17 |
|
zhar posted:I'm finally changing password manager from 1password to keepass (free!) and looking up good password practices it seems my master password isn't as secure as it should be. Or maybe it is as password databases have extra security to prevent brute force attacks, I don't know. What is the minimum strength the master password should be? I might just reuse my old one or make a new one but I don't want to overdo it as I'll be typing it a lot. Obviously I don't want to underdo it either. Is 4 random words enough? Keepass supports smart card or yubikeys as a auth option, make the text password an unintelligible string to be printed and put in a fireproof safe while using the physical token for daily use
|
# ? Mar 13, 2022 21:26 |
|
Powered Descent posted:This is all true, but keep in mind that upping the difficulty of the key derivation like this is going to run into diminishing returns pretty quickly. To use your example, let's say you increase the number of rounds so that every attempt takes thirty seconds instead of one. Great, brute-force attacks are now thirty times harder! ...but you're going to have to twiddle your thumbs for 30 seconds every single time you unlock it yourself. (If that's something you do once a week, that wait might be no big deal. If it's something you do twice an hour, you're going to want to punch your screen before long.) A much more convenient way to add difficulty is to simply use a longer password. Even if you limit yourself to letters and numbers, each extra random character you add to the password gives you something on the order of six more bits of entropy, which translates to an increase in brute-force difficulty of 64x -- more than twice as good as the big annoying 30-second wait. I don't doubt this and in fact i did some digging and found this calculator. There's a massive difference between length and iterations but I don't think you need a full thirty seconds to get enough benefit. I don't know what the default value of 100k iterations translates to in seconds to wait on a phone (if it's the 1password default not long enough I noticed the feature existed) but using a random 4 words from the smallest wordlist would supposedly cost $2,553,798 to break on average. I take this figure with a grain of salt (they do go through the maths and link to the original research) and I have no idea how to quantify how much the data from non-financial services gleaned from hacking all my accounts would be worth but I think I can confidently say I would be worth significantly less than this figure overall. Compare this with the $25.54 using only one iteration (I like to think I'm worth more than that at least). You'd need to use a length of 6 words with one iteration to beat the former figure (albeit significantly at $1,544,183,490.71) so a longer password is of course better (exponentially better) than more iterations but the combined total just needs to be sufficiently expensive. If I can get away with a little less finger fatigue with a 4 worder by having to wait a second I'll take it. SlowBloke posted:Keepass supports smart card or yubikeys as a auth option, make the text password an unintelligible string to be printed and put in a fireproof safe while using the physical token for daily use I'd never considered this. I don't know how much to trust these devices. If it got stolen along with a device containing the db wouldn't I be hosed? Would be convenient though.
|
# ? Mar 13, 2022 21:36 |
|
zhar posted:I'd never considered this. I don't know how much to trust these devices. If it got stolen along with a device containing the db wouldn't I be hosed? Would be convenient though. You can program the same seed on multiple yubikeys so if you lose one you won't get locked out. The database is not stored on the key, only the seed to generate the decryption code.
|
# ? Mar 13, 2022 22:08 |
|
SlowBloke posted:You can program the same seed on multiple yubikeys so if you lose one you won't get locked out. The database is not stored on the key, only the seed to generate the decryption code. I think they're asking if someone stole both the yubikey and the device with the database, if they would have access to all the passwords as well as all the accounts they go to. Which, yes, but that's not really a common threat model. And you could always hide the yubikey when you're not using it.
|
# ? Mar 13, 2022 22:29 |
|
And they’d still have to defeat any filesystem encryption as well.
|
# ? Mar 13, 2022 22:48 |
|
This is getting into Mossad/Not Mossad territory.
|
# ? Mar 13, 2022 23:33 |
|
KozmoNaut posted:Only if the attacker knows the pattern. Good thing patterns never get exposed in password leaks all the time
|
# ? Mar 14, 2022 00:53 |
|
Powered Descent posted:This is all true, but keep in mind that upping the difficulty of the key derivation like this is going to run into diminishing returns pretty quickly. To use your example, let's say you increase the number of rounds so that every attempt takes thirty seconds instead of one. Great, brute-force attacks are now thirty times harder! ...but you're going to have to twiddle your thumbs for 30 seconds every single time you unlock it yourself. (If that's something you do once a week, that wait might be no big deal. If it's something you do twice an hour, you're going to want to punch your screen before long.) A much more convenient way to add difficulty is to simply use a longer password. Oh absolutely, which is why I parenthetically said not to do that particular thing. +1 bit of entropy on the password is the same as doubling your KDF time. But there is a useful lesson to it, because long or extremely complex passwords are also a bit of a PITA. My master password is stupidly excruciating to type into the phone keyboard, it takes me like 20 seconds. I chose and memorized the password before I was syncing to a phone and didn't really thinking about it. It's a very strong password, keepass says 88 bits of entropy, go me! But after learning more about how it all works I would choose something that's faster to enter on the phone, and crank up the KDF difficulty. A happy median is a fine compromise. I haven't done anything about it because I don't need passwords on the phone very often, and not at all the last two years. :/ KozmoNaut posted:I have a few good randomly generated passwords that I can remember by the typing patterns, like for instance the general pattern is diagonally up and left, then down, then straight across. Jumps and upper/lowercase and symbols are of course scattered. Patterns based on any sort of keyboard progression like edcft6YHNJI(OL> or HUIjiojkl;nm,. are incredibly bad. First, it means you haven't "memorized" anything. If you go to france or encounter a dvorak keyboard you're hosed. Heck, I knew a guy who changed to a new model of ergo keyboard and had trouble with some passwords he only "knew" via muscle memory. Secondly. it's not hard to program a computer to generate keyboard-walk patterns. If it's a "pattern" based on the typography or arrangement of words, numbers, and symbols in a way you like that's ok. Maybe avoid stuff that 14 year olds do for their gamer handles like //|demon|slayer|\\ though. There's also stuff like the Schneier scheme where the idea is a personal idiosyncratic abbreviation of a memorable phrase. IMO the downside of that stuff is being harder to memorize & slower to type than a longer password of plain words. Especially on a phone screen where some symbols are 2 taps away.
|
# ? Mar 14, 2022 02:20 |
|
I just use 5 EFF dice ware words and call it a day, no symbols, spaces as separator
|
# ? Mar 14, 2022 03:30 |
|
Lol i was a page or two behind
|
# ? Mar 14, 2022 03:59 |
|
Mantle posted:Good thing patterns never get exposed in password leaks all the time If your unique pattern used for your KeePass password got leaked, that means your password got leaked, so you're hosed either way. Don't reuse patterns, duh. Klyith posted:
More like Mj#zVy%7 or something similar. That's up+left, down, right, up, right, but obviously skipping keys, I'm not suggesting a drat qwertyasdfgh password. Throw in some alternating left/right to make it easier to type. I don't have to worry about French or Dvorak keyboards etc., but if that's a concern for you, take your precautions.
|
# ? Mar 14, 2022 07:22 |
|
My CISO listed me as her delegate while she's on vacation this week so I'm sorry in advance for the imminent 0-day drops, as is tradition when I get put in charge :| Last time she did we got log4j lol I've learned nothing about agreeing to be a delegate.
|
# ? Mar 14, 2022 11:38 |
|
some kinda jackal posted:My CISO listed me as her delegate while she's on vacation this week so I'm sorry in advance for the imminent 0-day drops, as is tradition when I get put in charge :| What does she know that we don't know...
|
# ? Mar 14, 2022 16:46 |
|
some kinda jackal posted:I've learned nothing about agreeing to be a delegate. You've learned to let us all know so we can take some emergency leave this week, so that's at least something.
|
# ? Mar 14, 2022 16:49 |
|
RFC2324 posted:Am gonna say thats one thing I miss about the old stupid systems where the sysop could see your password: such comedy reading what people used
|
# ? Mar 14, 2022 19:23 |
|
y'all are way overthinking poo poo
|
# ? Mar 14, 2022 22:13 |
|
Achmed Jones posted:y'all are way overthinking poo poo Most people do. Which is fine, given that there's basically no functional cost imposed--remembering a 5 word phrase instead of a 4 word phrase is not gonna tax almost anyone. That said, AFAIK basically no criminal organization bothers to seriously try cracking hashed passwords unless you're Someone Special whose access might be specifically interesting. If you're running some sort of internet-exposed service on well known ports, then sure, maybe it gets scanned and someone tries to brute force it, but then the proper protection against that isn't having an extremely complex password, it's using rate limiting / banning on top of a not-trivial password, or preferably cert-based auth. Otherwise the real threat vector is password re-use allowing your Super Complex password to get pulled from some dumb company's un-encrypted / unsalted password db and sprayed back against all your other accounts. So as long as you don't use your master password anywhere else, then it's probably fine. Now if you are Someone Special who might get particular interest from bad people, then go ahead and use a bit more complex password just in case. But for everyone else there's no functional difference between a password that takes an hour to break on a high-end parallel-GPU-powered cracking machine and one that takes a million hours because no one's gonna bother to spend more than like, 1 minute on it, max. Just not worth the cost when there are plenty of people using default / trivial / repeat passwords out there. Same if someone steals your laptop/phone: if they can't pop it in a few minutes, your typical criminal is just gonna wipe the whole thing and sell it, or ditch it in the river. Just not worth the time otherwise. DrDork fucked around with this message at 22:55 on Mar 14, 2022 |
# ? Mar 14, 2022 22:53 |
|
Tldr: Mossad / Not Mossad threat analysis
|
# ? Mar 14, 2022 23:19 |
|
DrDork posted:That said, AFAIK basically no criminal organization bothers to seriously try cracking hashed passwords unless you're Someone Special whose access might be specifically interesting. Uh, did you mean to say a password manager database? Because regular hashed passwords get cracked all the time and people definitely spend time on it. But yeah, I can't imagine a criminal cracker who comes across a keepass database spending much time on it. Because at this point anyone who uses a password manager are at least minimally security conscious (or friends & family of a nerd who helped them set it up) and unlikely to use a easily-cracked password. Someone might run a dictionary of common passwords past it or something. Not bang on it for a week with a GPU farm. Volmarias posted:Tldr: Mossad / Not Mossad threat analysis I love that Mickens essay as much as anyone, but it's worth considering that he drastically underestimated the capabilities of not-Mossad, even at the time it was written. Also that several nation-states, including the actual Mossad, are now operating private enterprise pipelines that are criminal or insanely corrupt.
|
# ? Mar 15, 2022 03:27 |
Whether you call it MOSSAD or a private enterprise APT, it's still the same thing to me.
|
|
# ? Mar 15, 2022 13:00 |
|
BlankSystemDaemon posted:Whether you call it MOSSAD or a private enterprise APT, it's still the same thing to me.
|
# ? Mar 15, 2022 13:31 |
|
# apt install backdoor
|
# ? Mar 15, 2022 13:55 |
some kinda jackal posted:# apt install backdoor
|
|
# ? Mar 15, 2022 14:55 |
|
excuse you old man its # docker run backdoor:latest now
|
# ? Mar 15, 2022 14:56 |
|
|
# ? May 30, 2024 18:04 |
|
some kinda jackal posted:# apt install backdoor could not open lock file /srv/forums.somethingawful - (13: permission denied) CommieGIR fucked around with this message at 15:13 on Mar 15, 2022 |
# ? Mar 15, 2022 15:11 |