Hide Forgot
Description of problem: Version-Release number of selected component (if applicable): openswan-2.6.32-4.el6_1.1 How reproducible: always Steps to Reproduce: 1.Setup /etc/ipsec.conf with: conn testA auto=add authby=secret type=reject left=127.0.0.1 right=127.0.0.1 ike=3des-sha1 loopback=yes 2.service ipsec restart 3.ipsec auto --up testA Actual results: Packets aren't dropped and command ipsec auto --up testA has output: 030 "testA": cannot initiate an authby=never connection Expected results: Packets are dropped Additional info: Same behaviour has type=passthrough and type=reject
Sorry for mistake, ipsec.conf should be: conn testA auto=add authby=secret type=drop left=127.0.0.1 right=127.0.0.1 ike=3des-sha1 loopback=yes
Whatever you are trying to achieve, you have to do following: ipsec auto --route testA And then observe if it does whatever you want to achieve. In addition, why are you testing with openswan-2.6.32-4.el6_1.1 instead of openswan-2.6.32-6.el6?
Funcionality is ok with command ipsec auto --route testA with openswan-2.6.32-6.el6. But I find another problem with combination different types of connections. ipsec.conf look like this: conn testA auto=add authby=secret type=transport left=127.0.0.1 right=127.0.0.1 ike=3des-sha1 phase2=esp loopback=yes conn testB auto=add authby=secret type=passthrough left=127.0.0.1 right=127.0.0.1 ike=3des-sha1 leftprotoport=tcp/10000 loopback=yes It is important the order of calling ipsec auto --route testA and ipsec auto --route testB commands. 1. If ipsec auto --route testB called first and then ipsec auto --route testA, the behaviour is as I expecting (packets on port 10000 are without ipsec) 2. But if I call ipsec auto --route testA first and then ipsec auto --route testB (stricter rule). All packets are with ipsec (on port 10000 too). I am not sure that it is a bug or expected behaviour, but records in ip xfrm policy are same in both cases and in manual pages isn't describe that the order of calling ipsec auto --route is important.
Avesh, we have double-checked this and it turns out that all types of connection (ie. option 'type' in conn settings) works fine. But, it appears to an issue with an order in which different types are 'routed'. Please see BZ#742126 [1]. As this bug seems not be valid anymore, I am closing is as NOTABUG. [1] https://bugzilla.redhat.com/show_bug.cgi?id=742126