Talk:Online Certificate Status Protocol
|WikiProject Computing / Security||(Rated C-class)|
Article says that Safari supports OCSP but it needs to be enabled in Keychain access. However, I have this disabled in Keychain Access but Safari is still querying OCSP servers. I suspect the settings in Keychain Access are ignored by Safari, and Safari has been using OCSP for some time. -- Ch'marr (talk) 00:34, 31 August 2011 (UTC)
Who runs the OCSP servers? Is there more than one?
This piece of information would be very useful for people who are trying to gauge the merits of OCSP. Are OCSP servers more trustworthy than certificate authorities? If they are run by businesses, how do the businesses make money?
OCSP requests always go to the certificate authority that signed the certificate in question - those are the ones with the authority to revoke them
Okay, so, here we have some stuff. This is it. We are ready to go and make decisions. But the OCSP protocol is down. So what do we do?
A protocol is a recipe for procedures. A consortium somewhere, out there, has a protocol that they are not satisfied with. And for this reason I am being denied internet service and provided poor, harmful internet service.
My suggestion is that you provide (I'm guessing it's Mozilla Corps?) internet service without the OCSP protocol. Delete the text document describing the OSCP protocol specifications and procedures, and do not use it anymore. Then, go to the little switch thing that provides people internet, and turn __that__ one on.
My name is mmkstarr and my e-mail address (which I can't access at the moment, b/c your OCSP protocol doesn't work at all and so just give up on it) is firstname.lastname@example.org I am interested in hearing how things work, even if the mail doesn't reach me until after I'm not able to receive it. Which is how all mail works.
Another way to contact me is to drop pamphlets from helicopters. You could do that. In fact, I want to be contacted--so try that.
Further, if you simply have no one to __talk__ to, I recommend social media outlets, or objects, or other people.
Signing Off In Hopes Of A Silent Continuation,
So have you guys fixed the problem yet? Can I help in any way?
Are OCSP a privacy risk?
It seems like contacting an OCSP server might have privacy risks. First, it creates a record on-the-wire of every secure site a user connects to. Not only can the OCSP server maintain this log, but eavesdroppers as well. Additionally, the article mentions that it is typically an HTTP connection (not HTTPS).
- Q: What solution eliminates the info leak to the CA?
A: OCSP stapling with fallback to OCSP disabled.
- Q: Why is OCSP traffic typically HTTP and not HTTPS? -- the response is signed by the CA so, in theory, you can't really forge them -- the protocol has been partially broken see: http://www.blackhat.com/presentations/bh-usa-09/MARLINSPIKE/BHUSA09-Marlinspike-DefeatOCSP-PAPER2.pdf
A: Perhaps to avoid infinite loops where it's necessary to check the revocation status of a cert in order to check the invocation status of the same cert. --Noiseiron (talk) 22:48, 9 May 2016 (UTC)
- Q: Can a user select OCSP servers that he/she trusts?
A: The CA is the only actor that would be expected to have complete knowledge of what certificates it has issued and subsequently revoked and therefore how to respond to an OCSP request - at least until CAs implement blockchain PKI or until Certificate Transparency is ubiquitous. If the CA is not trusted with keeping private the list of sites we access, perhaps we shouldn't trust it with verifying the authenticity of the sites we are attempting to communicate with. --Noiseiron (talk) 22:48, 9 May 2016 (UTC)
- Can browsers be configured to connect to OCSP servers only via HTTPS? — Preceding unsigned comment added by 184.108.40.206 (talk) 13:57, 18 October 2011 (UTC)
Not sure a reference is very reliable
Currently refernce no. 5 ( "No, Don't Enable Revocation Checking". 19 April 2014. Retrieved 24 April 2014. https://www.imperialviolet.org/2014/04/19/revchecking.html) seems to have unreliable information. It starts off by referring to the Heartbleed bug as the "Heartbeat" bug. This error is rather more significant than a typo and makes one wonder if they are very well informed about the topic at all. I believe it should be removed as well as any information that was used from the article. 220.127.116.11 (talk) 17:58, 23 December 2015 (UTC)
Maybe mention DOS attacks using OCSP: https://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2016-6304 — Preceding unsigned comment added by FunnyDrink (talk • contribs) 13:49, 5 October 2016 (UTC)