I have a scenario where I need to connect to HTTPS site from my application (glassfish server or apache server). Let me illustrate via some example.
For Eg. I need to connect to https://stackoverflow.com from my application. Stackoverflow has chaining as below
If I try to validate from the client with all the 3 certificate( endsite cert,intermediate,cert and root cert) all works well.
The Question is, which is the right way of providing certificates from the client. If I just provide the *.stackexchange.com certificate from the client like below it works( So should I use enduser cert or root cert)
openssl s_client -CAfile stackexchangecom.crt -connect stackoverflow.com:443
So If understanding is correct then the webserver at stackoverflow.com has bundled the enduser certificate with Intermediate certificate and with the Root CA hence I get the valid connection.
Note : I am aware of the way SSL certificates work and also the chaining. I just want to know the best or the pragmatic practice for providing certificates for accessing the HTTPS
Related
I have multiple Intel NUCs out in the field that I use for displaying digital signage. Out of the thousands I have in the field, a couple complain with this error:
SSL certificate problem: self signed certificate in certificate chain. More details here: http://curl.haxx.se/docs/sslcerts.html, curl performs SSL certificate verification by default, using a "bundle" of Certificate Authority (CA) public keys (CA certs). If the default bundle file isn't adequate, you can specify an alternate file using the --cacert option. If this HTTPS server uses a certificate signed by a CA represented in the bundle, the certificate verification probably failed due to a problem with the certificate (it might be expired, or the name might not match the domain name in the URL). If you'd like to turn off curl's verification of the certificate, use the -k (or --insecure) option.
I've confirmed that the cert is not expired, and the domain name matches
These units are running on Debian 7
Could the network they are on cause this issue with some sort of firewall setting?
When you visit the website what certificate is shown in the web browser? In your curl bundle is that the same certificate which is supposed to be used for SSL encryption? I'd guess no. CA will sign your cert.pem so that web browsers will show your green lock thereby authenticating your website. You most likely have an issue on the back end with the configuration of your curl bundle. You need to make sure that your server is using the CA cert not a self signed certificate like ssl-cert-snakeoil.pem for example.
Essentially your website should be using a static IPv4 address. As far as a network firewall stopping an SSL handshake from happening that may be possible, I have seen it happen on specific ports for example port 22 for ssh connections may be blocked at the network gateway for inbound traffic on a client side computer attempting a connection to a server. The SYN/ACK https://tools.ietf.org/rfc/rfc793.txt TCP handshake may time out in that type of network fire wall situation. However since you are getting an explicit response from your server about a self signed cert a firewall issue does not seem to be the problem.
I created self signed certificate and is being used in my express project. The problem is when I do
curl https://<domain>
doesn't and only works with -k option, it doesn't use the certificate to encrypt its content. I don't see how that is effective since that is similar to http itself.
Also from what I understand in CA signing request is that browsers like chrome they are preinstalled with the information trusted CAs in the web such as Verisign. So how does curl work in this way, does it also know before hand like in chrome regarding who are the trusted CAs?
Also in my project the app communicates with the ec2 machine does it need client side certificate for https ?
Yes, curl has a pointer to a list of trusted Certificate Authority keys. You can override it with the --capath flag.
During SSL handshake, the browser downloads any intermediate certificate from the host web server using the URL provided if needed. I believe browser comes with the pre-installed certificates from public CAs having only the public key of the root certificate.
1) When calling a https url using a standalone java program [https://xyz.com ..which is using Verisign certificate], i do not need to add that Certificate to any truststore since its root public is already available in jdk's cacerts truststore file. Is this correct statement?
2) When i run the same program from application server, it requires to add all the intermediate certificate to server truststore individually. Why this works in different way.
If the trust chain for the servers certificate is: root-intermediate#1-intermediate#2-server and the client (browser) has root as trusted CA in its CA store, it needs a way to verify the servers certificate by checking the complete chain up to the root. And because the client usually has no knowledge of the intermediate CAs the server needs to provide them.
Sometimes it seems to work w/o providing these intermediate CAs. First, the browsers usually cache the intermediate CAs they got and thus if intermediate#2 is the same as already seen by another server the verification will succeed, but only for the clients who visited the other server before :(
Another way is to provide a URL inside a certificate, where the issuer certificate can be downloaded, e.g. server could provide a link to the certificate for intermediate#2, intermediate#2 could provide a link to intermediate#1. In this case the client could download the missing certificates. But, this features is not universally adopted, e.g. some browsers might provide it but SSL libraries outside of the browsers usually don't.
We have an Azure web role deployed that uses HTTPS. We upload a certificate to azure and shortly after the portal refreshes and two more certificate appear. This is not a wild cart certificate and maybe this is standard behavior, but I haven't seen it before.
The original certificate is named something like:
subdomain.domain.com
The three certificates that appear are named like so:
VeriSign Class 3 Public Primary Certification Authority - G5
Class 3 Public Primary Certification Authority
VeriSign Class 3 International Server CA - G3
Are the 3 certificates I mentioned normally generated or is this an issue I should be looking into?
We have a similar deployment that has an ssl, but does not generate these extra certificates. This is what triggered our concern and has me asking why ...?
When you enabled HTTPS endpoint in any web application and bind SSL certificate to it, the certificate bind to HTTPS endpoint is could be a single certificate or it could be a chain and
it is depend on several factors as below:
When the certificate is created as self signed ROOT then it will have only one certificate in the chain. This certificate can not be validated to have SSL tunnel because there is no other part to verify it and that why it is called self signed root
When you buy certificate from a reputed CA (Certificate Authority) in almost all cases you will get 3 (or more) certificates:
2.1. Root Certificate : This certificate is helps to create a SSL tunnel between two machines using PKI security Infrastructure.
2.2. Intermediate Certificate -> This is to create a chain with multiple certs as if needed
2.3. Domain Certificate -> This is for your *.domainname.com or domainname.com
Here is an example of chained SSL certificate at https://mail.google.com
And all of these certificate are chained into one single PFX (if private key embedded into certs) or CER (without any Private Key) so when you deploy only ONE PFX cert, you see the chain is open and all certificates are listed.
If you browser your url and open the certificate view through browser, you will see exactly same chain as you could see in your portal and you can also verify the certificate thumbprint as well to match.
I have a server with an SSL certificate and clients with SSL cetificates, all are signed by the same CA, and the CA is trusted on the server and clients as a root authority.
However none of the clients I have tried (iphone, chrome, explorer) will send the client certificate when the server requests it, even though they all verify the server certificate fine. They all claim not to have the certificate.
When I look at the client certificate in the certificate/profile/store they all claim it is verified and legit, and all the certificates verify fine using openssl etc.
Does an SSL client certificate need a specific name or other details so that client browsers will know when to use it? Kinda of like how a server certificate has to specify the exact domain as the name on the certificate?
The server is an apache2.2 server, but I don't think its a problem on the servers end.
Client certificates must have appropriate Key Usage and Extended Key Usage extensions set. What is "appropriate" in your case, I don't know for sure but the number of variants in Key Usage is small and you can check various values yourself.
You surely haven't forgot about the private key which should be accessible on the client side, have you?
There's a summary of the key usage and extended key usage extensions to use in this document:
http://www.mozilla.org/projects/security/pki/nss/tech-notes/tn3.html
(Although it is for NSS, it should apply to other products.)
You could also check whether the list of accepted CAs sent by the server is configured properly. This can be seen using openssl s_client -connect the.host.name:443, for example.