<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Comments inline.<br>
    <br>
    <div class="moz-cite-prefix">On 4/8/14, 11:16 AM, Josh Gachnang
      wrote:<br>
    </div>
    <blockquote
cite="mid:CAOrc4CyJA18vVhFP64r5tmsv5TMM4ZNppS25+YMWeS7HRRdvUA@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <blockquote class="gmail_quote" style="margin:0px 0px 0px
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><span
            style="font-family:arial,sans-serif;font-size:13px">I'm more
            accustomed to using PDUs for this type of thing. I.e., a<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">power
            strip you can ssh into or hit via a web API to toggle power
            to<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">individual
            ports.</span><br
            style="font-family:arial,sans-serif;font-size:13px">
          <span style="font-family:arial,sans-serif;font-size:13px">Machines
            are configured to power up on power restore, plus PXE boot.<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">You have
            less control than with IPMI -- all you can do is toggle
            power<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">to the
            outlet -- but it works well, even for some desktop machines
            I<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">have in
            a lab.</span><br
            style="font-family:arial,sans-serif;font-size:13px">
          <span style="font-family:arial,sans-serif;font-size:13px">I
            don't have a compelling need, but I've often wondered if
            such a<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">driver
            would be useful. I can imagine it also being useful if
            people<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">want to
            power up non-compute stuff, though that's probably not a top<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">priority
            right now.</span></blockquote>
        <div><br>
        </div>
        <div>I believe someone was talking about this yesterday in the
          meeting. It would be very possible to write an IPMI driver
          (possibly being renamed for this reason) to control the power
          of a node via a PDU. You could then plug that into the agent
          driver as the power driver to create something like
          AgentAndPDUDriver. The current agent driver doesn't do
          anything with IPMI except set boot device. The inability to
          set boot device would be the biggest issue with a PDU driver
          as far as I can see, but that's not insurmountable.</div>
      </div>
    </blockquote>
    <br>
    +1<br>
    <br>
    The /agent itself/ being a power driver, as suggested earlier, seems
    like it wouldn't work well though. Honestly, any situation that'd
    require running the agent on the tenant, should be out of scope.
    This is explicitly a ramdisk agent, and it should be optimized to
    run in a ramdisk, not on resources assigned to a tenant.<br>
    <br>
    <blockquote
cite="mid:CAOrc4CyJA18vVhFP64r5tmsv5TMM4ZNppS25+YMWeS7HRRdvUA@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div><br>
        </div>
        <blockquote class="gmail_quote" style="margin:0px 0px 0px
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><span
            style="font-family:arial,sans-serif;font-size:13px">How much
            hardware information do we intend to store in Ironic? (Note<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">that I'm
            genuinely asking this, not challenging your assertion.) It<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">seems
            reasonable, but I think there's a lot of hardware
            information<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">that
            could be useful (say, lspci output, per-processor flags,
            etc.),<br>
          </span><span
            style="font-family:arial,sans-serif;font-size:13px">but
            stuffing it all in extra[] seems kind of messy.</span></blockquote>
        <div><br>
        </div>
        <div>Right now the hardware manager on the agent is pluggable,
          so what we're storing is currently "whatever you want!". I
          think in our current iteration, it is just the MACs of the
          NICs. We haven't fully fleshed this out yet.</div>
      </div>
      <div class="gmail_extra"><br clear="all">
      </div>
    </blockquote>
    <br>
    Jim and I are working on patches to the agent to send up more
    information, including ram/cpu/block devices and some information
    from DMI (like serial numbers) and lldp in order to help populate
    neutron.<br>
    <br>
    I'm of the opinion that generally more is better, as long as it's
    long-lived information (such as RAM/CPUs/etc) that doesn't change
    except in cases of an explicit maintenance.<br>
    <br>
    <br>
    -Jay<br>
    <blockquote
cite="mid:CAOrc4CyJA18vVhFP64r5tmsv5TMM4ZNppS25+YMWeS7HRRdvUA@mail.gmail.com"
      type="cite">
      <div class="gmail_extra">
        <div>
          <div dir="ltr">---<br>
            Josh Gachnang<br>
            <div><span
style="border-collapse:collapse;font-family:arial,sans-serif;font-size:13px">Tech
                Blog: ServerCobra.com, @ServerCobra
                <div>
                  Github.com/PCsForEducation</div>
              </span></div>
          </div>
        </div>
        <br>
        <br>
        <div class="gmail_quote">On Tue, Apr 8, 2014 at 10:46 AM, Matt
          Wagner <span dir="ltr"><<a moz-do-not-send="true"
              href="mailto:matt.wagner@redhat.com" target="_blank">matt.wagner@redhat.com</a>></span>
          wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            On 08/04/14 14:04 +0400, Vladimir Kozhukalov wrote:<br>
            <snip>
            <div class=""><br>
              <blockquote class="gmail_quote" style="margin:0 0 0
                .8ex;border-left:1px #ccc solid;padding-left:1ex">
                0) There are a plenty of old hardware which does not
                have IPMI/ILO at all.<br>
                How Ironic is supposed to power them off and on? Ssh?
                But Ironic is not<br>
                supposed to interact with host OS.<br>
              </blockquote>
              <br>
            </div>
            I'm more accustomed to using PDUs for this type of thing.
            I.e., a<br>
            power strip you can ssh into or hit via a web API to toggle
            power to<br>
            individual ports.<br>
            <br>
            Machines are configured to power up on power restore, plus
            PXE boot.<br>
            You have less control than with IPMI -- all you can do is
            toggle power<br>
            to the outlet -- but it works well, even for some desktop
            machines I<br>
            have in a lab.<br>
            <br>
            I don't have a compelling need, but I've often wondered if
            such a<br>
            driver would be useful. I can imagine it also being useful
            if people<br>
            want to power up non-compute stuff, though that's probably
            not a top<br>
            priority right now.
            <div class=""><br>
              <br>
              <blockquote class="gmail_quote" style="margin:0 0 0
                .8ex;border-left:1px #ccc solid;padding-left:1ex">
                1) We agreed that Ironic is that place where we can
                store hardware info<br>
                ('extra' field in node model). But many modern hardware
                configurations<br>
                support hot pluggable hard drives, CPUs, and even
                memory. How Ironic will<br>
                know that hardware configuration is changed? Does it
                need to know about<br>
                hardware changes at all? Is it supposed that some
                monitoring agent (NOT<br>
                ironic agent) will be used for that? But if we already
                have discovering<br>
                extension in Ironic agent, then it sounds rational to
                use this extension<br>
                for monitoring as well. Right?<br>
              </blockquote>
              <br>
            </div>
            How much hardware information do we intend to store in
            Ironic? (Note<br>
            that I'm genuinely asking this, not challenging your
            assertion.) It<br>
            seems reasonable, but I think there's a lot of hardware
            information<br>
            that could be useful (say, lspci output, per-processor
            flags, etc.),<br>
            but stuffing it all in extra[] seems kind of messy.<br>
            <br>
            I don't have an overall answer for this question; I'm
            curious myself.<span class="HOEnZb"><font color="#888888"><br>
                <br>
                -- Matt</font></span>
            <div class="HOEnZb">
              <div class="h5"><br>
                <br>
                _______________________________________________<br>
                OpenStack-dev mailing list<br>
                <a moz-do-not-send="true"
                  href="mailto:OpenStack-dev@lists.openstack.org"
                  target="_blank">OpenStack-dev@lists.openstack.org</a><br>
                <a moz-do-not-send="true"
                  href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev"
                  target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
              </div>
            </div>
          </blockquote>
        </div>
        <br>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
OpenStack-dev mailing list
<a class="moz-txt-link-abbreviated" href="mailto:OpenStack-dev@lists.openstack.org">OpenStack-dev@lists.openstack.org</a>
<a class="moz-txt-link-freetext" href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a>
</pre>
    </blockquote>
    <br>
  </body>
</html>