1

This may be a stupid question, but I'll ask anyway. I have been reading quite a bit about the Heart Bleed vulnerability, and I'm curious as to whether it only affects web applications or whether OVPN connections secured with certs produced with OpenSSL would potentially be affected.

I have attempted to scan an OVPN Server I know to be secured with OpenSSL certificates at the website http://filippo.io/Heartbleed/, however I get an error in regards to StartTLS. I don't have the ability to execute the Python script where I am, but I'm wondering if this connection could be vulnerable to this attack.

DKNUCKLES
  • 9,237
  • 2
  • 37
  • 47

2 Answers2

4

In short, OpenVPN using buggy OpenSSL versions are indeed affected. OpenSSL versions older than 1.0.1 and newer than (and including) 1.0.1g are safe. Versions in between may be safe iff the vendor applied a patch. Note that there are other SSL libraries that can be used with OpenVPN, namely PolarSSL.

Long story, OpenVPN often uses UDP instead of TCP. Existing public tools are mostly concentrated on TLS over TCP. So for that reason alone, you cannot just enter your OpenVPN server running on UDP into a random website and expect a sane response.

Second, OpenVPN uses its own protocol on top of TCP. TLS run atop of that protocol. Looking at a packet capture for OpenVPN (over UDP), I can see that the Heartbeat extension is advertised.

If you have not updated your OpenVPN client now, hurry! If you manage the OpenVPN server and have still not considered updating your OpenSSL libraries, have you been sleeping for a whole day?!

Lekensteyn
  • 5,898
  • 5
  • 37
  • 62
  • Neither the client nor server appear to be vulnerable. If certs were issued with OpenSSL do they need to be revoked and re-issued? – DKNUCKLES Apr 09 '14 at 15:02
  • @DKNUCKLES What makes you think that the **OpenVPN** service is not vulnerable? Certificate creation has nothing to do with transport layer encryption. – Lekensteyn Apr 09 '14 at 15:04
  • Because the vendor has stated that it's not vulnerable. http://forum.mikrotik.com/viewtopic.php?p=420218#p420218 – DKNUCKLES Apr 09 '14 at 15:09
  • @DKNUCKLES Ah, but you miss the crucial part of that quote: *ALL prior RouterOS releases (6.11 and older) are not affected by this vulnerability as **older OpenSSL library where used**.* Not all OpenSSL versions are affected, see the [advisory](http://www.openssl.org/news/secadv_20140407.txt). – Lekensteyn Apr 09 '14 at 15:11
  • How am I missing any part of that quote? If older OpenSSL libraries are used which do not appear on the advisory, then what's the risk? – DKNUCKLES Apr 09 '14 at 15:15
  • @DKNUCKLES You did not mention your OpenSSL library version, so I assumed it was a recent, vulnerable version. I have edited the answer to clarify that it only applies to some versions. – Lekensteyn Apr 09 '14 at 15:17
1

Just to be explicit, the only reason to revoke and re-issue an SSL cert is if you were running a recent, affected version of OpenSSL which was used in a service that exposed HeartBeating and you fear that someone could have exploited HeartBleed to acquire your private SSL key. There is no fundamental security issue with SSL certs issued with vulnerable versions of OpenSSL.

If you were running an older version of OpenSSL, then you don't have a problem.

If you were using OpenSSL in such a way that did not expose HeartBeating, then you don't have a problem.

If you patched and are confident that no one attempted to exploit you, or if they did they didn't get the proverbial keys to the castle, then you probably don't have a problem... but there's no way to be absolutely sure about this one. It's likely that you're OK, but there's no good metric by which we can judge actual exposure.

Jason
  • 1,907
  • 2
  • 9
  • 15
  • Thank you - I don't think I completely comprehended the vulnerability when I posted, but this information is very useful. Thanks! – DKNUCKLES Apr 09 '14 at 15:56