
Nadav, Thank you for working on this but we have one more issue with name resolution. I checked the last job you triggered and I noticed that vm migration failed due to similar issue between the hosts. Here is a piece of custom logs that you added: 2017-04-30 14:23:35,675-0400 INFO (Reactor thread) [ProtocolDetector.SSLHandshakeDispatcher] subject: ((('organizationName', u'Test'),), (('commonName', u'lago-basic-suite-master-host0'),)), key: organizationName, value: Test (sslutils:241) 2017-04-30 14:23:35,675-0400 INFO (Reactor thread) [ProtocolDetector.SSLHandshakeDispatcher] subject: ((('organizationName', u'Test'),), (('commonName', u'lago-basic-suite-master-host0'),)), key: commonName, value: lago-basic-suite-master-host0 (sslutils:241) 2017-04-30 14:23:35,676-0400 INFO (Reactor thread) [ProtocolDetector.SSLHandshakeDispatcher] src_addr: ::ffff:192.168.201.2, cn_addr: lago-basic-suite-master-host0 (sslutils:262) 2017-04-30 14:23:35,676-0400 INFO (Reactor thread) [ProtocolDetector.SSLHandshakeDispatcher] src_addr_extracted: 192.168.201.2, cn_addr_extracted: lago-basic-suite-master-host0 (sslutils:266) 2017-04-30 14:23:35,677-0400 INFO (Reactor thread) [ProtocolDetector.SSLHandshakeDispatcher] socket.gethostbyadd(src_addr)[0]: lago-basic-suite-master-host0.lago.local (sslutils:268) 2017-04-30 14:23:35,678-0400 INFO (Reactor thread) [ProtocolDetector.SSLHandshakeDispatcher] compare ::ffff:192.168.201.2, lago-basic-suite-master-host0, res: False (sslutils:244) 2017-04-30 14:23:35,678-0400 ERROR (Reactor thread) [ProtocolDetector.SSLHandshakeDispatcher] peer certificate does not match host name (sslutils:226) It looks like the engine issued certificate for 'lago-basic-suite-master-host0' but we resolve 192.168.201.2 to 'lago-basic-suite-master-host0.lago.local'. Can we fix it as well? Thanks, Piotr On Sun, Apr 30, 2017 at 7:42 PM, Piotr Kliczewski <pkliczew@redhat.com> wrote:
Wow, great.
Thank you!
30 kwi 2017 19:40 "Nadav Goldin" <ngoldin@redhat.com> napisał(a):
Ok, I think the issue was the unqualified domain name. The certificate was generated(as before for 'engine') without the domain name, i.e. 'lago-basic-suite-master-engine', on VDSM side it resolved the IP to the address 'lago-basic-suite-master-engine.lago.local' and then failed comparing it to the unqualified one. I assume this is the expected behaviour, though not sure(as you can easily resolve 'lago-basic-suite-master-engine' to 'lago-basic-suite-master-engine.lago.local' on the hosts). It should be fixed in [1], just ran OST manual with the same debugging patch applied on top of yours, and at least add_hosts passed.
[1] https://gerrit.ovirt.org/#/c/76225/10 [2] http://jenkins.ovirt.org/job/ovirt-system-tests_manual/338/console
On Sun, Apr 30, 2017 at 7:50 PM, Piotr Kliczewski <pkliczew@redhat.com> wrote:
Sure, will take look later today.
30 kwi 2017 18:47 "Nadav Goldin" <ngoldin@redhat.com> napisał(a):
Thanks for the explanation.
I added some more debugging messages on top of your patch, could you please take a look at [1] and tell me what do you expect to resolve differently for this to work?
[1]
http://jenkins.ovirt.org/job/ovirt-system-tests_manual/337/artifact/exported...