Re: [ovirt-users] New problem with hosted-engine during "Configuring the management bridge"
I have moved on. I honestly can't say what fixed this problem, but not
seeing it any more.
Then I ran into a problem with NIC naming that apuimedo (and jvandewege)
helped me through in IRC, but hosted-engine completes successfully now.
I can't configure my VM engine network for some reason, but I'll look
into that more deeply later.
What's the recommended procedure for self-hosted using F20 engine when
it comes to the network configuration wizard.
Leave it alone?
Set for static config and configure with FQDN and address on main host
network?
Thanks,
Bob
On 05/15/2014 11:56 AM, Dan Kenigsberg wrote:
On Thu, May 15, 2014 at 04:20:09PM +0200, Sandro Bonazzola wrote:
> Il 15/05/2014 16:09, Bob Doolittle ha scritto:
>> On 05/15/2014 09:59 AM, Sandro Bonazzola wrote:
>>> Il 15/05/2014 15:55, Bob Doolittle ha scritto:
>>>> If I run "vdsm-tool service-reload multipathd" by hand then I
see the log I'd expect:
>>>> reload multipathd.service
>>> If this command works, hosted-engine setup should not fail on it.
>>> The setup doesn't touch anything related to multipathd.
>> Sorry but that is not strictly correct.
>> You can see the code in multipath.py (setupMultipath()) that writes a new
multipath.conf file just before it tries to restart the service.
>> It looks like it's trying to modify the scsi_id_path.
>>
>> So when I run it by hand, I am using the original multipath.conf file. When it is
run during hosted-engine setup, it is with a modified version.
>> My efforts in modifying the python have been to try to capture that modified
version, so that I can exercise it by hand and see what's wrong.
> If something is configuring it, it must be vdsm-tool:
As I've noted ealier on this thread, that's not correct, unfortunately:
Bug 1076531 - vdsm overwrites multipath.conf at every startup
> In hosted-engine setup code there's a call to "vdsm-tool configure
--force".
However, it's possible to copy the vdsm-sactioned multipath.conf from
one host to another, and try `systemctl reload multipathd.service`.
Please look into journalctl for hints on why it has failed.