<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <p>Hi,</p>
    <p>Have not measured it as we have been running this way for years
      now and haven't experienced any problems with "transport endpoint
      is not connected” with this setup.</p>
    <p>We used the default options "BONDING_OPTS='mode=6 miimon=100'"</p>
    <blockquote>
      <dl class="variablelist compact">
        <dt><span class="term"> <code class="literal">miimon=<em
                class="replaceable">time_in_milliseconds</em> </code> </span></dt>
        <dd>
          <div class="para"> Specifies (in milliseconds) how often MII
            link monitoring occurs. This is useful if high availability
            is required because MII is used to verify that the NIC is
            active. </div>
        </dd>
      </dl>
    </blockquote>
    <p><br>
    </p>
    <div class="moz-cite-prefix">On 2/25/19 2:22 PM, Martin Toth wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:9B82193A-15EF-43F6-882B-BA8E7862770A@gmail.com">
      <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
      How long does it take to your devices (using mode 5 or 6, ALB is
      prefered for GlusterFS) to take-over the MAC? This can result in
      your error -  "transport endpoint is not connected” - there are
      some timeouts within gluster set by default.
      <div class="">I am using LACP and it works without any problem.
        Can you share your mode 5 / 6 configuration ?</div>
      <div class=""><br class="">
      </div>
      <div class="">Thanks.</div>
      <div class="">Martin<br class="">
        <div class="">
          <div><br class="">
            <blockquote type="cite" class="">
              <div class="">On 25 Feb 2019, at 13:44, Jorick Astrego
                &lt;<a href="mailto:jorick@netbulae.eu" class=""
                  moz-do-not-send="true">jorick@netbulae.eu</a>&gt;
                wrote:</div>
              <br class="Apple-interchange-newline">
              <div class="">
                <meta http-equiv="Content-Type" content="text/html;
                  charset=UTF-8" class="">
                <div text="#000000" bgcolor="#FFFFFF" class="">
                  <p class="">Hi,</p>
                  <p class="">Well no, mode 5 and mode 6 also have fault
                    tollerance and don't need any special switch config.</p>
                  <p class="">Quick google search:</p>
                  <p class=""><a class="moz-txt-link-freetext"
href="https://serverfault.com/questions/734246/does-balance-alb-and-balance-tlb-support-fault-tolerance"
                      moz-do-not-send="true">https://serverfault.com/questions/734246/does-balance-alb-and-balance-tlb-support-fault-tolerance</a><br
                      class="">
                  </p>
                  <blockquote class="">
                    <p class="">Bonding Mode 5 (balance-tlb) works by
                      looking at all the devices in the bond, and
                      sending out the slave with the least current
                      traffic load. Traffic is only received by one
                      slave (the "primary slave"). If a slave is lost,
                      that slave is not considered for transmission, so
                      this mode is fault-tolerant.</p>
                    <p class="">Bonding Mode 6 (balance-alb) works as
                      above, except incoming ARP requests are
                      intercepted by the bonding driver, and the bonding
                      driver generates ARP replies so that external
                      hosts are tricked into sending their traffic into
                      one of the other bonding slaves instead of the
                      primary slave. If many hosts in the same broadcast
                      domain contact the bond, then traffic should
                      balance roughly evenly into all slaves.</p>
                    <p class="">If a slave is lost in Mode 6, then it
                      may take some time for a remote host to time out
                      its ARP table entry and send a new ARP request. A
                      TCP or SCTP retransmission tents to lead into ARP
                      request fairly quickly, but a UDP datagram does
                      not, and will rely on the usual ARP table refresh.
                      So Mode 6 <em class="">is</em> fault tolerant,
                      but convergence on slave loss may take some time
                      depending on the Layer 4 protocol used.</p>
                    <p class="">If you are worried about fast fault
                      tolerance, then consider using Mode 4 (802.3ad aka
                      LACP) which negotiates link aggregation between
                      the bond and the switch, and constantly updates
                      the link status between the aggregation partners.
                      Mode 4 also has configurable load balance hashing
                      so is better for in-order delivery of TCP streams
                      compared to Mode 5 or Mode 6.</p>
                  </blockquote>
                  <p class=""><a class="moz-txt-link-freetext"
                      href="https://wiki.linuxfoundation.org/networking/bonding"
                      moz-do-not-send="true">https://wiki.linuxfoundation.org/networking/bonding</a><br
                      class="">
                  </p>
                  <ul class="">
                    <li class="level2">
                      <div class="li"> <strong class=""> balance-tlb or
                          5</strong><br class="">
                        Adaptive transmit load balancing: channel
                        bonding that does not require any special switch
                        support. The outgoing traffic is distributed
                        according to the current load (computed relative
                        to the speed) on each slave. Incoming traffic is
                        received by the current slave. <b class="">If
                          the receiving slave fails, another slave takes
                          over the MAC address of the failed receiving
                          slave.</b></div>
                      <ul class="">
                        <li class="level3">
                          <div class="li"> Prerequisite:</div>
                          <ol class="">
                            <li class="level4">
                              <div class="li"> Ethtool support in the
                                base drivers for retrieving the speed of
                                each slave.</div>
                            </li>
                          </ol>
                        </li>
                      </ul>
                    </li>
                    <li class="level2">
                      <div class="li"> <strong class=""> balance-alb or
                          6 </strong><br class="">
                        Adaptive load balancing: <b class="">includes
                          balance-tlb plus receive load balancing</b>
                        (rlb) for IPV4 traffic, and does not require any
                        special switch support. The receive load
                        balancing is achieved by ARP negotiation.</div>
                      <ul class="">
                        <li class="level3">
                          <div class="li"> The bonding driver intercepts
                            the ARP Replies sent by the local system on
                            their way out and overwrites the source
                            hardware address with the unique hardware
                            address of one of the slaves in the bond
                            such that different peers use different
                            hardware addresses for the server.</div>
                        </li>
                        <li class="level3">
                          <div class="li"> Receive traffic from
                            connections created by the server is also
                            balanced. When the local system sends an ARP
                            Request the bonding driver copies and saves
                            the peer's IP information from the ARP
                            packet.</div>
                        </li>
                        <li class="level3">
                          <div class="li"> When the ARP Reply arrives
                            from the peer, its hardware address is
                            retrieved and the bonding driver initiates
                            an ARP reply to this peer assigning it to
                            one of the slaves in the bond.</div>
                        </li>
                        <li class="level3">
                          <div class="li"> A problematic outcome of
                            using ARP negotiation for balancing is that
                            each time that an ARP request is broadcast
                            it uses the hardware address of the bond.
                            Hence, peers learn the hardware address of
                            the bond and the balancing of receive
                            traffic collapses to the current slave. This
                            is handled by sending updates (ARP Replies)
                            to all the peers with their individually
                            assigned hardware address such that the
                            traffic is redistributed. Receive traffic is
                            also redistributed when a new slave is added
                            to the bond and when an inactive slave is
                            re-activated. The receive load is
                            distributed sequentially (round robin) among
                            the group of highest speed slaves in the
                            bond.</div>
                        </li>
                        <li class="level3">
                          <div class="li"> When a link is reconnected or
                            a new slave joins the bond the receive
                            traffic is redistributed among all active
                            slaves in the bond by initiating ARP Replies
                            with the selected mac address to each of the
                            clients. The updelay parameter (detailed
                            below) must be set to a value equal or
                            greater than the switch's forwarding delay
                            so that the ARP Replies sent to the peers
                            will not be blocked by the switch.</div>
                        </li>
                      </ul>
                    </li>
                  </ul>
                  <div class="moz-cite-prefix">On 2/25/19 1:16 PM,
                    Martin Toth wrote:<br class="">
                  </div>
                  <blockquote type="cite"
                    cite="mid:ED57C951-1D9B-47F2-967E-A2F1FB0382AE@gmail.com"
                    class="">
                    <meta http-equiv="Content-Type" content="text/html;
                      charset=UTF-8" class="">
                    Hi Alex,
                    <div class=""><br class="">
                    </div>
                    <div class="">you have to use bond mode 4 (LACP -
                      802.3ad) in order to achieve redundancy of
                      cables/ports/switches. I suppose this is what you
                      want.</div>
                    <div class=""><br class="">
                    </div>
                    <div class="">BR,</div>
                    <div class="">Martin</div>
                    <div class="">
                      <div class="">
                        <div class=""><br class="">
                          <blockquote type="cite" class="">
                            <div class="">On 25 Feb 2019, at 11:43, Alex
                              K &lt;<a
                                href="mailto:rightkicktech@gmail.com"
                                class="" moz-do-not-send="true">rightkicktech@gmail.com</a>&gt;
                              wrote:</div>
                            <br class="Apple-interchange-newline">
                            <div class="">
                              <div dir="ltr" class="">
                                <div class="">Hi All, <br class="">
                                </div>
                                <div class=""><br class="">
                                </div>
                                <div class="">I was asking if it is
                                  possible to have the two separate
                                  cables connected to two different
                                  physical switched. When trying mode6
                                  or mode1 in this setup gluster was
                                  refusing to start the volumes, giving
                                  me "transport endpoint is not
                                  connected". <br class="">
                                </div>
                                <div class=""><br class="">
                                </div>
                                <div class="">server1: cable1
                                  ---------------- switch1
                                  --------------------- server2: cable1<br
                                    class="">
                                </div>
                                <div class="">                              
                                               |<br class="">
                                </div>
                                <div class="">server1: cable2
                                  ---------------- switch2
                                  --------------------- server2: cable2<br
                                    class="">
                                </div>
                                <div class=""><br class="">
                                </div>
                                <div class="">Both switches are
                                  connected with each other also. This
                                  is done to achieve redundancy for the
                                  switches. <br class="">
                                </div>
                                <div class="">When disconnecting cable2
                                  from both servers, then gluster was
                                  happy. <br class="">
                                </div>
                                <div class="">What could be the problem?<br
                                    class="">
                                </div>
                                <div class=""><br class="">
                                </div>
                                <div class="">Thanx,</div>
                                <div class="">Alex<br class="">
                                </div>
                                <div class=""><br class="">
                                </div>
                              </div>
                              <br class="">
                              <div class="gmail_quote">
                                <div dir="ltr" class="gmail_attr">On
                                  Mon, Feb 25, 2019 at 11:32 AM Jorick
                                  Astrego &lt;<a
                                    href="mailto:jorick@netbulae.eu"
                                    class="" moz-do-not-send="true">jorick@netbulae.eu</a>&gt;
                                  wrote:<br class="">
                                </div>
                                <blockquote class="gmail_quote"
                                  style="margin:0px 0px&#xA; 0px
                                  0.8ex;border-left:1px solid&#xA;
                                  rgb(204,204,204);padding-left:1ex">
                                  <div bgcolor="#FFFFFF" class="">
                                    <p class="">Hi,</p>
                                    <p class="">We use bonding mode 6
                                      (balance-alb) for GlusterFS
                                      traffic</p>
                                    <p class=""><a
                                        class="gmail-m_-151486401419277343moz-txt-link-freetext"
href="https://access.redhat.com/documentation/en-us/red_hat_gluster_storage/3.4/html/administration_guide/network4"
                                        target="_blank"
                                        moz-do-not-send="true">https://access.redhat.com/documentation/en-us/red_hat_gluster_storage/3.4/html/administration_guide/network4</a><br
                                        class="">
                                    </p>
                                    <blockquote class="">
                                      <p class="">Preferred bonding mode
                                        for Red Hat Gluster Storage
                                        client is mode 6 (balance-alb),
                                        this allows client to transmit
                                        writes in parallel on separate
                                        NICs much of the time. <br
                                          class="">
                                      </p>
                                    </blockquote>
                                    <p class="">Regards,</p>
                                    <p class="">Jorick Astrego<br
                                        class="">
                                    </p>
                                    <div
                                      class="gmail-m_-151486401419277343moz-cite-prefix">On
                                      2/25/19 5:41 AM, Dmitry Melekhov
                                      wrote:<br class="">
                                    </div>
                                    <blockquote type="cite" class="">
                                      <div
                                        class="gmail-m_-151486401419277343moz-cite-prefix">23.02.2019
                                        19:54, Alex K пишет:<br class="">
                                      </div>
                                      <blockquote type="cite" class="">
                                        <div dir="ltr" class="">
                                          <div class="">Hi all, <br
                                              class="">
                                          </div>
                                          <div class=""><br class="">
                                          </div>
                                          <div class="">I have a replica
                                            3 setup where each server
                                            was configured with a dual
                                            interfaces in mode 6
                                            bonding. All cables were
                                            connected to one common
                                            network switch. <br
                                              class="">
                                          </div>
                                          <div class=""><br class="">
                                          </div>
                                          <div class="">To add
                                            redundancy to the switch,
                                            and avoid being a single
                                            point of failure, I
                                            connected each second cable
                                            of each server to a second
                                            switch. This turned out to
                                            not function as gluster was
                                            refusing to start the volume
                                            logging "transport endpoint
                                            is disconnected" although
                                            all nodes were able to reach
                                            each other (ping) in the
                                            storage network. I switched
                                            the mode to mode 1
                                            (active/passive) and
                                            initially it worked but
                                            following a reboot of all
                                            cluster same issue appeared.
                                            Gluster is not starting the
                                            volumes. <br class="">
                                          </div>
                                          <div class=""><br class="">
                                          </div>
                                          <div class="">Isn't
                                            active/passive supposed to
                                            work like that? Can one have
                                            such redundant network setup
                                            or are there any other
                                            recommended approaches?</div>
                                          <div class=""><br class="">
                                          </div>
                                        </div>
                                      </blockquote>
                                      <p class=""><br class="">
                                      </p>
                                      <p class="">Yes, we use lacp, I
                                        guess this is mode 4 ( we use
                                        teamd ), it is, no doubt, best
                                        way.</p>
                                      <p class=""><br class="">
                                      </p>
                                      <blockquote type="cite" class="">
                                        <div dir="ltr" class="">
                                          <div class="">Thanx, <br
                                              class="">
                                          </div>
                                          <div class="">Alex<br class="">
                                          </div>
                                        </div>
                                        <br class="">
                                        <fieldset
                                          class="gmail-m_-151486401419277343mimeAttachmentHeader"></fieldset>
                                        <pre class="gmail-m_-151486401419277343moz-quote-pre">_______________________________________________
Gluster-users mailing list
<a class="gmail-m_-151486401419277343moz-txt-link-abbreviated" href="mailto:Gluster-users@gluster.org" target="_blank" moz-do-not-send="true">Gluster-users@gluster.org</a>
<a class="gmail-m_-151486401419277343moz-txt-link-freetext" href="https://lists.gluster.org/mailman/listinfo/gluster-users" target="_blank" moz-do-not-send="true">https://lists.gluster.org/mailman/listinfo/gluster-users</a></pre>
                                      </blockquote>
                                      <p class=""><br class="">
                                      </p>
                                      <br class="">
                                      <fieldset
                                        class="gmail-m_-151486401419277343mimeAttachmentHeader"></fieldset>
                                      <pre class="gmail-m_-151486401419277343moz-quote-pre">_______________________________________________
Gluster-users mailing list
<a class="gmail-m_-151486401419277343moz-txt-link-abbreviated" href="mailto:Gluster-users@gluster.org" target="_blank" moz-do-not-send="true">Gluster-users@gluster.org</a>
<a class="gmail-m_-151486401419277343moz-txt-link-freetext" href="https://lists.gluster.org/mailman/listinfo/gluster-users" target="_blank" moz-do-not-send="true">https://lists.gluster.org/mailman/listinfo/gluster-users</a></pre>
                                    </blockquote>
                                    <br class="">
                                    <br class="">
                                    <br class="">
                                    <br class="">
                                    <span style="color:rgb(96,76,120)"
                                      class=""><font class=""><span
                                          class="" lang="NL">Met
                                          vriendelijke groet, With kind
                                          regards,<br class="">
                                          <br class="">
                                          Jorick Astrego<br class="">
                                        </span></font></span><b
                                      style="color:rgb(96,76,120)"
                                      class=""><br class="">
                                      Netbulae Virtualization Experts </b><br
                                      class="">
                                    <hr
                                      style="border-color:rgb(204,204,204)&#xA;
                                      currentcolor
                                      currentcolor;border-style:solid&#xA;
                                      none none;border-width:1px medium
                                      medium" class="">
                                    <table style="width:522px" class="">
                                      <tbody class="">
                                        <tr class="">
                                          <td
                                            style="width:130px;font-size:10px"
                                            class="">Tel: 053 20 30 270</td>
                                          <td
                                            style="width:130px;font-size:10px"
                                            class=""><a
                                              href="mailto:info@netbulae.eu"
                                              target="_blank" class=""
                                              moz-do-not-send="true">info@netbulae.eu</a></td>
                                          <td
                                            style="width:130px;font-size:10px"
                                            class="">Staalsteden 4-3A</td>
                                          <td
                                            style="width:130px;font-size:10px"
                                            class="">KvK 08198180</td>
                                        </tr>
                                        <tr class="">
                                          <td
                                            style="width:130px;font-size:10px"
                                            class="">Fax: 053 20 30 271</td>
                                          <td
                                            style="width:130px;font-size:10px"
                                            class=""><a
                                              href="http://www.netbulae.eu/"
                                              target="_blank" class=""
                                              moz-do-not-send="true">www.netbulae.eu</a></td>
                                          <td
                                            style="width:130px;font-size:10px"
                                            class="">7547 TA Enschede</td>
                                          <td
                                            style="width:130px;font-size:10px"
                                            class="">BTW NL821234584B01</td>
                                        </tr>
                                      </tbody>
                                    </table>
                                    <br class="">
                                    <hr
                                      style="border-color:rgb(204,204,204)&#xA;
                                      currentcolor
                                      currentcolor;border-style:solid&#xA;
                                      none none;border-width:1px medium
                                      medium" class=""><br class="">
                                  </div>
_______________________________________________<br class="">
                                  Gluster-users mailing list<br class="">
                                  <a
                                    href="mailto:Gluster-users@gluster.org"
                                    target="_blank" class=""
                                    moz-do-not-send="true">Gluster-users@gluster.org</a><br
                                    class="">
                                  <a
                                    href="https://lists.gluster.org/mailman/listinfo/gluster-users"
                                    rel="noreferrer" target="_blank"
                                    class="" moz-do-not-send="true">https://lists.gluster.org/mailman/listinfo/gluster-users</a></blockquote>
                              </div>
_______________________________________________<br class="">
                              Gluster-users mailing list<br class="">
                              <a href="mailto:Gluster-users@gluster.org"
                                class="" moz-do-not-send="true">Gluster-users@gluster.org</a><br
                                class="">
                              <a class="moz-txt-link-freetext"
                                href="https://lists.gluster.org/mailman/listinfo/gluster-users"
                                moz-do-not-send="true">https://lists.gluster.org/mailman/listinfo/gluster-users</a></div>
                          </blockquote>
                        </div>
                        <br class="">
                      </div>
                    </div>
                    <br class="">
                    <fieldset class="mimeAttachmentHeader"></fieldset>
                    <pre class="moz-quote-pre" wrap="">_______________________________________________
Gluster-users mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Gluster-users@gluster.org" moz-do-not-send="true">Gluster-users@gluster.org</a>
<a class="moz-txt-link-freetext" href="https://lists.gluster.org/mailman/listinfo/gluster-users" moz-do-not-send="true">https://lists.gluster.org/mailman/listinfo/gluster-users</a></pre>
                  </blockquote>
                  <br class="">
                  <br class="">
                  <br class="">
                  <br class="">
                  <span style="color:#604c78;" class=""><font class=""><span
                        style="mso-fareast-language:en-gb;" class=""
                        lang="NL">Met vriendelijke groet, With kind
                        regards,<br class="">
                        <br class="">
                        Jorick Astrego<br class="">
                      </span></font></span><b style="color:#604c78"
                    class=""><br class="">
                    Netbulae Virtualization Experts </b><br class="">
                  <hr style="border:none;border-top:1px solid #ccc;"
                    class="">
                  <table style="width: 522px" class="">
                    <tbody class="">
                      <tr class="">
                        <td style="width: 130px;font-size: 10px"
                          class="">Tel: 053 20 30 270</td>
                        <td style="width: 130px;font-size: 10px"
                          class=""><a href="mailto:info@netbulae.eu"
                            class="" moz-do-not-send="true">info@netbulae.eu</a></td>
                        <td style="width: 130px;font-size: 10px"
                          class="">Staalsteden 4-3A</td>
                        <td style="width: 130px;font-size: 10px"
                          class="">KvK 08198180</td>
                      </tr>
                      <tr class="">
                        <td style="width: 130px;font-size: 10px"
                          class="">Fax: 053 20 30 271</td>
                        <td style="width: 130px;font-size: 10px"
                          class=""><a href="http://www.netbulae.eu"
                            class="" moz-do-not-send="true">www.netbulae.eu</a></td>
                        <td style="width: 130px;font-size: 10px"
                          class="">7547 TA Enschede</td>
                        <td style="width: 130px;font-size: 10px"
                          class="">BTW NL821234584B01</td>
                      </tr>
                    </tbody>
                  </table>
                  <br class="">
                  <hr style="border:none;border-top:1px solid #ccc;"
                    class=""><br class="">
                </div>
                _______________________________________________<br
                  class="">
                Gluster-users mailing list<br class="">
                <a href="mailto:Gluster-users@gluster.org" class=""
                  moz-do-not-send="true">Gluster-users@gluster.org</a><br
                  class="">
                <a class="moz-txt-link-freetext" href="https://lists.gluster.org/mailman/listinfo/gluster-users">https://lists.gluster.org/mailman/listinfo/gluster-users</a></div>
            </blockquote>
          </div>
          <br class="">
        </div>
      </div>
    </blockquote>
  <BR />
<BR />
<b style="color:#604c78"></b><br><br><span style="color:#604c78;"><font color="000000"><span style="mso-fareast-language:en-gb;" lang="NL">Met vriendelijke groet, With kind regards,<br><br>Jorick Astrego<br></span></font></span><b style="color:#604c78"><br>Netbulae Virtualization Experts </b><br><hr style="border:none;border-top:1px solid #ccc;"><table style="width: 522px"><tbody><tr><td style="width: 130px;font-size: 10px">Tel:  053 20 30 270</td>    <td style="width: 130px;font-size: 10px">info@netbulae.eu</td>    <td style="width: 130px;font-size: 10px">Staalsteden 4-3A</td>    <td style="width: 130px;font-size: 10px">KvK 08198180</td></tr><tr>    <td style="width: 130px;font-size: 10px">Fax: 053 20 30 271</td>    <td style="width: 130px;font-size: 10px">www.netbulae.eu</td>    <td style="width: 130px;font-size: 10px">7547 TA Enschede</td>    <td style="width: 130px;font-size: 10px">BTW NL821234584B01</td></tr></tbody></table><br><hr style="border:none;border-top:1px solid #ccc;"><BR />
</body>
</html>