Thanks a lot for you effort, I'm glad that you were able to upgrade successfully although we were not able to find the cause for the issue :-(

On Thu, Jul 21, 2016 at 2:30 PM, <nicolas@devels.es> wrote:
So I gave it another try and this time it worked without any issue (with 4.0.1.1 version). Strange, maybe the first upgrade failure left system in a weird state? Anyhow almost everything ([1]) is working fine now. Thanks for the help!

  [1]: https://bugzilla.redhat.com/show_bug.cgi?id=1358737

​Adding Tomas about this one
 

El 2016-07-20 20:23, Martin Perina escribió:
On Wed, Jul 20, 2016 at 6:18 PM, Nicolás <nicolas@devels.es> wrote:

El 20/07/16 a las 16:45, Martin Perina escribió:

On Wed, Jul 20, 2016 at 4:44 PM, Nicolás <nicolas@devels.es> wrote:

Hi Martin,

Actually, up until now we had that cert configured in httpd and in
websocket proxy. Seems that now in 4.0.x it's not enough, as opening
the https://fqdn [1] complains about the cert not being imported in
the key chain.

​Yes, there's an updated procedure on using external CA in 4.0,
for details please take a look at Doc Text in

https://bugzilla.redhat.com/show_bug.cgi?id=1336838 [2]
​  
So I imported it via keytool, but I don't want to use it in the
engine <-> VDSM communication.

​Hmm, so that would imply that we have some issue with existing
internal enigne CA during upgrade ...

The strange thing is that we test upgrades a lot but so far we
haven't seen any issues which will broke

SSL setup between engine and VDSM. You said that you had to
downgrade back to 3.6.7 (so unfortunately for us we cannot
investigate your nonworking setup more), but how did you do that?

Removing all engine packages and configuration​, installing back
3.6.7 packaging and restoring configuration form backup?

I'm asking to know what changed in your setup between not working
4.0 and working 3.6.7 ...

 Indeed, those are the steps I followed to the point.

 To add more strangeness, previously to upgrading this oVirt
infrastructure, we upgraded another one that we have (also using own
cert, a different one but from the same CA) and everything went
smoothly. And what's more, previously to upgrading the engine that
failed, I created a copy of that engine machine in a sandbox
environment to see if upgrade process would or not success, and it
worked perfectly.

 The only difference between the sandbox and the real machine's
process was that when upgrading the real one, the first time I run
"engine-setup" it failed because 'systemd' reported PostgreSQL as it
was not running (actually it was, thougg), so everything rolled back.
I had to kill the PostgreSQL process, start it again with systemctl
and then run "engine-setup", where the process completed successfully
but the SSL issue appeared. Not sure if this rollback could have
shattered the whole thing...

 Anyhow, tomorrow I'm going to create another copy of the engine
machine to a sandbox environment and try again. If it works I'll cross
my fingers and give another try on the real machine...

 Thanks!

​Thanks a lot for you effort. I will try to perform same upgrade
tomorrow in my test env.
​ 

Thanks

Martin

Thanks!
En 20/7/2016 2:48 p. m., Martin Perina <mperina@redhat.com>
escribió:

Hi,

sorry for late response, I overlook your reply :-(

​I looked at your logs and it seems to me that​ there's SSL
error when engine tries to contact VDSM.

​You have mentioned that your are using your own custom CA. ​Are
you using it only for HTTPS certificate or do you want to use it
also for Engine <-> VDSM communication?


Martin Perina

 

On Wed, Jul 20, 2016 at 9:18 AM, <nicolas@devels.es> wrote:
Any hints about this?

El 2016-07-13 11:13, nicolas@devels.es escribió:
Hi,

Unfortunately, upgrading to 4.0.1RC didn't solve the problem.
Actually, the error changed to 'General SSLEngine problem', but the
result was the same, like this:

2016-07-13 09:52:22,010 INFO
[org.ovirt.vdsm.jsonrpc.client.reactors.ReactorClient] (SSL Stomp
Reactor) [] Connecting to /10.X.X.X
2016-07-13 09:52:22,018 ERROR
[org.ovirt.vdsm.jsonrpc.client.reactors.Reactor] (SSL Stomp
Reactor)
[] Unable to process messages: General SSLEngine problem

It's worth mentioning that we're using our own SSL certificates
(not
self-signed), and I imported the combined certificate into the
/etc/pki/ovirt-engine/.truststore key file. Not sure if related,
but
just in case.

I had to downgrade to 3.6.7. I'm attaching requested logs, if you
need
anything else don't hesitate to ask.

Regards.

El 2016-07-13 09:45, Martin Perina escribió:
Hi,

could you please share also vdsm.log from your hosts and also
server.log and setup logs from /var/log/ovirt-engine/setup
directory?

Thanks

Martin Perina

On Wed, Jul 13, 2016 at 10:36 AM, <nicolas@devels.es> wrote:

Hi,

We upgraded from 3.6.6 to 4.0.0 and we have a big issue since the
engine cannot connect to hosts. In the logs all we see is this
error:

    ERROR [org.ovirt.vdsm.jsonrpc.client.reactors.Reactor] (SSL
Stomp Reactor) [] Unable to process messages

I'm attaching full logs.

Could someone help please?

Thanks.
_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users [3] [1]

Links:
------
[1] http://lists.ovirt.org/mailman/listinfo/users [3]

 _______________________________________________
 Users mailing list
 Users@ovirt.org
 http://lists.ovirt.org/mailman/listinfo/users [3]



Links:
------
[1] https://fqdn
[2] https://bugzilla.redhat.com/show_bug.cgi?id=1336838
[3] http://lists.ovirt.org/mailman/listinfo/users