On 29/01/16 10:31, Christian Heimes wrote:
On 2016-01-28 19:56, Roderick Johnstone wrote:
On 28/01/16 13:39, Christian Heimes wrote:
On 2016-01-28 13:51, Roderick Johnstone wrote:
Hi
My netapp filer is happily doing ldap over ssl lookups for account
information to my RHEL 6.7 testing ipa server
(ipa-server-3.0.0-47.el6_7.1.x86_64).
However, when I switch the filer to use my RHEL 7.2 ipa server
(ipa-server-4.2.0-15.el7_2.3.x86_64) the lookup doesn't work.
In the dirsrv log file I see entries like this:
[28/Jan/2016:09:17:45 +0000] conn=1338 fd=112 slot=112 SSL connection
from xxx.xxx.xxx.xxx to yyy.yyy.yy.yyy
[28/Jan/2016:09:17:45 +0000] conn=1338 op=-1 fd=112 closed - Cannot
communicate securely with peer: no common encryption algorithm(s).
(xxx.xxx.xxx.xxx is the filer ip address and yyy.yyy.yyy.yyy is the ipa
server ip address).
Looking in the ldap directory for fields with cipher in the name shows a
very different set of nssslenabledciphers between the two ipa-server
versions.
I wonder if this might be the issue?
Can the ldap server tell me what ciphers its being requested to use by
the filer?
Yes, it looks like it is the issue. The supported cipher suites were
hardened a while ago. The ticket
https://fedorahosted.org/freeipa/ticket/4395 contains more information.
During the TLS handshake the client sends a list of supported cipher
suites to the server. The server also has a list of supported cipher
suites. But the server never sends this list to the client. Instead it
picks one common cipher suite (usually the most secure) from the common
set of cipher suites.
I don't know if you can get 389 DS to print the cipher suites. But you
can snoop the ciper suites from the TLS handshake with wireshark or
tshark. The handshake isnt't encrypted and can be captures on either the
host or the server.
# tshark -Vx -Y "ssl.handshake.ciphersuites" -i YOUR_INTERFACE tcp port
ldaps
Christian
Thanks Christian. Thats really helpful.
Now I have a list of ciphers being asked for and I found that the ldap
server logs which ciphers its using when it starts up file
/var/log/dirsrv/slapd-<domain>/error. There isn't any overlap.
I noticed that there is a setting in the
dn: cn=encryption,cn=config
allowWeakCipher: off
and
nsSSL3Ciphers: +all
and found some documentation on this here:
http://directory.fedoraproject.org/docs/389ds/design/nss-cipher-design.html
So, maybe I could add one (or several) of the required ciphers to
nsSSL3Ciphers or possibly as a last resort set allowWeakCipher: on?
Hi Roderick,
I highly recommend against lowering the settings. Weak ciphers are
broken and insecure ciphers, some even with NULL encryption or no
authentication. At best weak ciphers may (!) protect your against a
passive sniffer and incompetent attacker. They won't protect you against
a serious attack.
Are you able to reconfigure or update the client? Does the client even
speak TLS 1.0 to the server or is it restricted to SSLv2 and SSLv3?
If you show me the complete handshake, I can give you further advice.
The handshake output of tshark starts like this:
Secure Sockets Layer
SSL Record Layer: Handshake Protocol: Client Hello
Content Type: Handshake (22)
Version: TLS 1.0 (0x0301)
Christian
Christian
I don't think we have much control over the available client ciphers. We
are running the latest Data OnTap version for our natapps so we have
what we have. The netapp can do TLSv1 though.
We do have firewalling on the ipa servers so that will help until one of
our trusted networks is compromised!
I'll send you the handshake output from tshark off list.
Thanks
Roderick
--
Manage your subscription for the Freeipa-users mailing list:
https://www.redhat.com/mailman/listinfo/freeipa-users
Go to http://freeipa.org for more info on the project
