[ovirt-users] Prereqs for Self Hosted Engine Hyper Converged Gluster Support

[Michael DePaulo]

Hi Didi,

1st of all, note that I am trying this on my home network. This is not
a production environment by any means.

On Mon, Aug 3, 2015 at 1:52 AM, Yedidyah Bar David <didi at redhat.com> wrote:
> On Mon, Aug 3, 2015 at 6:44 AM, Michael DePaulo <mikedep333 at gmail.com> wrote:
>> Hi,
>>
>> I am trying to follow these instructions on 3.6 alpha 3:
>> http://www.ovirt.org/Features/Self_Hosted_Engine_Hyper_Converged_Gluster_Support
>>
>> The 1st time I ran ovirt-hosted-engine-setup (which was unsuccessful),
>
> What's exactly "unsuccessful"? Did it continue beyond the "closeup" stage?

I ran a total of 25 runs yesterday due to multiple issues that I ran
into. The reason why the setup failed seemed to be that the hosted
engine VM did not reboot and start up the engine quickly enough, so it
failed during health.py.
2015-08-02 14:03:26 ERROR otopi.context context._executeMethod:164
Failed to execute stage 'Closing up': timed out

I eventually worked around this by increasing the time.sleep() call to
a larger value in the installed health.py, and then recompiling the
.pyc and .pyo files.
https://github.com/oVirt/ovirt-hosted-engine-setup/blob/c6bc631b81c241c2bf5b5313997e7e575874ac8f/src/plugins/ovirt-hosted-engine-setup/engine/health.py#L161

I think I experienced a bug whereby the while loop is only running 2
or 3 times, but I do not see a bug in the code.

Anyway, I planned to send a separate email about this issue, hopefully
after I have investigated the possible bug with the loop and prepared
a patch.

>> I saw the prompt:
>>
>>     Do you want to configure this host for providing GlusterFS
>> storage? (Yes, No)[No]:
>>
>> Now I no longer see it.
>
> Please check/post setup logs (of all relevant runs), from
> /var/log/ovirt-hosted-engine-setup

This log shows the fact that I was not prompted to configure storage,
and it shows the "timed out" failure during health.py.
http://pastebin.com/5x3Xhss7

>
>>
>> Must gluster not have any peers or volumes already? Is there some
>> other prereq that I am missing?
>
> Not sure, adding Sandro.
>
> Note that the official support for HC was postponed to 4.0 [1], but some
> parts were already implemented and should work.
>
> [1] https://bugzilla.redhat.com/show_bug.cgi?id=1175354
>
> Best,
> --
> Didi

Thank you,
-Mike