<div dir="ltr"><div class="gmail_default" style="font-size:large"><br></div><div class="gmail_extra"><br><div class="gmail_quote">On Sat, Sep 30, 2017 at 2:51 PM, Dan Yasny <span dir="ltr"><<a href="mailto:dyasny@gmail.com" target="_blank">dyasny@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="auto">The power management command is sent by the engine via a proxy host. That means you need at least one more host to act as proxy. The engine itself doesn't need to access the bmc network directly. Just like the engine needs no access to the atorage network to perform storage manipulations. <div dir="auto"><br></div><div dir="auto">I think in some recent versions fencing by the engine was introduced, but I don't have a setup in front of me to verify.</div></div></blockquote><div><br></div><div><div style="font-size:large" class="gmail_default">No, this is an open RFE which we didn't apply yet and as far as I know there is no plan for that in the near future.</div><div style="font-size:large" class="gmail_default">So, currently there is a need for an additional host that will serve as a proxy to the fencing operation.</div><div style="font-size:large" class="gmail_default">The main reason for the existence of the proxy host is when a host that has power management configure becomes non-responsive</div><div style="font-size:large" class="gmail_default">In that case we might restart the host via its power management card <br></div><div style="font-size:large" class="gmail_default">But, since this host may have some HA VMs that will be migrated to another host before the problematic host is rebooted, we have to implement a <br></div><div style="font-size:large" class="gmail_default">shutdown->wait for OFF status from PM card->migrate HA VMs->start->wait for ON status from PM card <br></div><div style="font-size:large" class="gmail_default">This is done in order to prevent running a VM on two hosts which will cause corruption (brain split)</div><div style="font-size:large" class="gmail_default"><br></div><br></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div class="gmail-HOEnZb"><div class="gmail-h5"><div class="gmail_extra"><br><div class="gmail_quote">On Sep 29, 2017 11:13 PM, "~Stack~" <<a href="mailto:i.am.stack@gmail.com" target="_blank">i.am.stack@gmail.com</a>> wrote:<br type="attribution"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On 09/29/2017 05:31 PM, Dan Yasny wrote:<br>
> You need more than one host for power management<br>
><br>
<br>
Seriously?? I didn't see anything like that in the docs...Maybe I just<br>
missed it.<br>
<br>
Also, why wouldn't it still validate? It should still be able to talk to<br>
the interface over the BMC/IPMI network. The fact that I can run the<br>
equivalent test on the command line makes it seem like it should at<br>
least be able to check via the test. Obviously, it would be silly for it<br>
to try to manage itself, but it could at least verify that the<br>
configuration is valid, right?<br>
<br>
I have more hosts that I'm going to add, I was just hoping to do those<br>
via the Foreman integration instead of manually building them. Since I'm<br>
not quite ready for that, I will just build a second host on Monday and<br>
report back.<br>
<br>
Thanks for the feedback. I would have never guess that as a possibility. :-)<br>
<br>
~Stack~<br>
<br>
</blockquote></div></div>
</div></div><br>______________________________<wbr>_________________<br>
Users mailing list<br>
<a href="mailto:Users@ovirt.org">Users@ovirt.org</a><br>
<a href="http://lists.ovirt.org/mailman/listinfo/users" rel="noreferrer" target="_blank">http://lists.ovirt.org/<wbr>mailman/listinfo/users</a><br>
<br></blockquote></div><br></div></div>