<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">On 08/14/2017 03:19 PM, Sriram wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAMvdjuri0fpdzd-ATfO-M99bw=L3-iYcqTcrtmoV2bkJNOe8VA@mail.gmail.com">
      <div dir="ltr">Yes, I had precreated the script file with the
        required permission. 
        <div>
          <div><br>
          </div>
          <div>[root@<b>node1</b> alerts]# ls -l
            /usr/share/pacemaker/alert_file.sh</div>
          <div>-rwxr-xr-x. 1 root root 4140 Aug 14 01:51
            /usr/share/pacemaker/alert_file.sh</div>
        </div>
        <div>
          <div>
            <div> [root@<b>node2</b> alerts]# ls -l
              /usr/share/pacemaker/alert_file.sh</div>
            <div>-rwxr-xr-x. 1 root root 4139 Aug 14 01:51
              /usr/share/pacemaker/alert_file.sh</div>
          </div>
          <div>[root@<b>node3</b> alerts]# ls -l
            /usr/share/pacemaker/alert_file.sh</div>
          <div>-rwxr-xr-x. 1 root root 4139 Aug 14 01:51
            /usr/share/pacemaker/alert_file.sh</div>
        </div>
        <div><br>
        </div>
        <div>Later I observed that user "hacluster" is not able to
          create the log file under <span style="font-size:12.8px">/usr/share/pacemaker/</span><wbr
            style="font-size:12.8px"><span style="font-size:12.8px">alert_file.log.</span></div>
        <div>I am sorry, I should have observed this in the log before
          posting the query. Then I gave the path as
          /tmp/alert_file.log, it is able to create now.</div>
        <div>
          <div>Thanks for pointing it out.</div>
        </div>
        <div><br>
        </div>
        <div>I have one more clarification,</div>
        <div><br>
        </div>
        <div>if the resource is running in node2,</div>
        <div>[root@node2 tmp]# pcs resource<br>
        </div>
        <div>
          <div> TRR    (ocf::heartbeat:TimingRedundancyRA):    Started
            node2</div>
        </div>
        <div><br>
        </div>
        <div>And I executed the below command to make it standby.</div>
        <div>[root@node2 tmp] # pcs node standby node2</div>
        <div><br>
        </div>
        <div>Resource shifted to node3, because of higher location
          constraint.</div>
        <div>
          <div>
            <div>[root@node2 tmp]# pcs resource</div>
            <div> TRR    (ocf::heartbeat:TimingRedundancyRA):    Started
              node3.</div>
          </div>
        </div>
        <div><br>
        </div>
        <div><br>
        </div>
        <div>I got the log file created under node2(resource stopped)
          and node3(resource started). </div>
        <div><br>
        </div>
        <div>Node1 was not notified about the resource shift, I mean no
          log file was created there.</div>
        <div>Its because alerts are designed to notify the external
          agents about the cluster events. Its not for internal
          notifications.</div>
        <div><br>
        </div>
        <div>Is my understanding correct ?</div>
      </div>
    </blockquote>
    <br>
    Quite simple: crmd of node1 just didn't have anything to do with
    shifting the resource<br>
    from node2 -> node3. There is no additional information passed
    between the nodes<br>
    just to create a full set of notifications on every node. If you
    want to have a full log<br>
    (or whatever you altert-agent is doing) in one place this would be
    up to your alert-agent.<br>
    <br>
    Regards,<br>
    Klaus<br>
    <br>
    <blockquote type="cite"
cite="mid:CAMvdjuri0fpdzd-ATfO-M99bw=L3-iYcqTcrtmoV2bkJNOe8VA@mail.gmail.com">
      <div dir="ltr">
        <div> </div>
        <div>Regards,<br>
        </div>
        <div>Sriram.</div>
        <div><br>
        </div>
        <div><br>
        </div>
      </div>
      <div class="gmail_extra"><br>
        <div class="gmail_quote">On Mon, Aug 14, 2017 at 5:42 PM, Klaus
          Wenninger <span dir="ltr"><<a
              href="mailto:kwenning@redhat.com" target="_blank"
              moz-do-not-send="true">kwenning@redhat.com</a>></span>
          wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div text="#000000" bgcolor="#FFFFFF">
              <div>
                <div class="h5">
                  <div class="m_-4802990479422366187moz-cite-prefix">On
                    08/14/2017 12:32 PM, Sriram wrote:<br>
                  </div>
                  <blockquote type="cite">
                    <div dir="ltr">Hi Ken,
                      <div><br>
                      </div>
                      <div>I used the alerts as well, seems to be not
                        working.</div>
                      <div><br>
                      </div>
                      <div>Please check the below configuration</div>
                      <div>
                        <div>[root@node1 alerts]# pcs config show</div>
                        <div>Cluster Name:</div>
                        <div>Corosync Nodes:</div>
                        <div>Pacemaker Nodes:</div>
                        <div> node1 node2 node3</div>
                        <div><br>
                        </div>
                        <div>Resources:</div>
                        <div> Resource: TRR (class=ocf
                          provider=heartbeat type=TimingRedundancyRA)</div>
                        <div>  Operations: start interval=0s timeout=60s
                          (TRR-start-interval-0s)</div>
                        <div>              stop interval=0s timeout=20s
                          (TRR-stop-interval-0s)</div>
                        <div>              monitor interval=10
                          timeout=20 (TRR-monitor-interval-10)</div>
                        <div><br>
                        </div>
                        <div>Stonith Devices:</div>
                        <div>Fencing Levels:</div>
                        <div><br>
                        </div>
                        <div>Location Constraints:</div>
                        <div>  Resource: TRR</div>
                        <div>    Enabled on: node1 (score:100)
                          (id:location-TRR-node1-100)</div>
                        <div>    Enabled on: node2 (score:200)
                          (id:location-TRR-node2-200)</div>
                        <div>    Enabled on: node3 (score:300)
                          (id:location-TRR-node3-300)</div>
                        <div>Ordering Constraints:</div>
                        <div>Colocation Constraints:</div>
                        <div>Ticket Constraints:</div>
                        <div><br>
                        </div>
                        <div>Alerts:</div>
                        <div> Alert: alert_file
                          (path=/usr/share/pacemaker/<wbr>alert_file.sh)</div>
                        <div>  Options: debug_exec_order=false</div>
                        <div>  Meta options: timeout=15s</div>
                        <div>  Recipients:</div>
                        <div>   Recipient: recipient_alert_file_id
                          (value=/usr/share/pacemaker/<wbr>alert_file.log)</div>
                      </div>
                    </div>
                  </blockquote>
                  <br>
                </div>
              </div>
              Did you pre-create the file with proper rights? Be aware
              that the alert-agent<br>
              is called as user hacluster.<span class=""><br>
                <br>
                <blockquote type="cite">
                  <div dir="ltr">
                    <div>
                      <div><br>
                      </div>
                      <div>Resources Defaults:</div>
                      <div> resource-stickiness: INFINITY</div>
                      <div>Operations Defaults:</div>
                      <div> No defaults set</div>
                      <div><br>
                      </div>
                      <div>Cluster Properties:</div>
                      <div> cluster-infrastructure: corosync</div>
                      <div> dc-version: 1.1.15-11.el7_3.4-e174ec8</div>
                      <div> default-action-timeout: 240</div>
                      <div> have-watchdog: false</div>
                      <div> no-quorum-policy: ignore</div>
                      <div> placement-strategy: balanced</div>
                      <div> stonith-enabled: false</div>
                      <div> symmetric-cluster: false</div>
                      <div><br>
                      </div>
                      <div>Quorum:</div>
                      <div>  Options:</div>
                    </div>
                    <div><br>
                    </div>
                    <div><br>
                    </div>
                    <div>/usr/share/pacemaker/alert_<wbr>file.sh does
                      not get called whenever I trigger a scenario for
                      failover.<br>
                    </div>
                    <div>Please let me know if I m missing anything. <br>
                    </div>
                  </div>
                </blockquote>
                <br>
              </span> Do you get any logs - like for startup of
              resources - or nothing at all?<br>
              <br>
              Regards,<br>
              Klaus
              <div>
                <div class="h5"><br>
                  <br>
                  <blockquote type="cite">
                    <div dir="ltr">
                      <div class="gmail_extra"><br>
                      </div>
                      <div class="gmail_extra"><br>
                      </div>
                      <div class="gmail_extra">Regards,</div>
                      <div class="gmail_extra">Sriram.</div>
                      <div class="gmail_extra"><br>
                        <div class="gmail_quote">On Tue, Aug 8, 2017 at
                          8:29 PM, Ken Gaillot <span dir="ltr"><<a
                              href="mailto:kgaillot@redhat.com"
                              target="_blank" moz-do-not-send="true">kgaillot@redhat.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
                              class="m_-4802990479422366187gmail-HOEnZb">
                              <div
                                class="m_-4802990479422366187gmail-h5">On
                                Tue, 2017-08-08 at 17:40 +0530, Sriram
                                wrote:<br>
                                > Hi Ulrich,<br>
                                ><br>
                                ><br>
                                > Please see inline.<br>
                                ><br>
                                > On Tue, Aug 8, 2017 at 2:01 PM,
                                Ulrich Windl<br>
                                > <<a
                                  href="mailto:Ulrich.Windl@rz.uni-regensburg.de"
                                  target="_blank" moz-do-not-send="true">Ulrich.Windl@rz.uni-regensbur<wbr>g.de</a>>
                                wrote:<br>
                                >         >>> Sriram <<a
                                  href="mailto:sriram.ec@gmail.com"
                                  target="_blank" moz-do-not-send="true">sriram.ec@gmail.com</a>>
                                schrieb am <a href="tel:08.08.2017"
                                  value="+498082017" target="_blank"
                                  moz-do-not-send="true">08.08.2017</a>
                                um<br>
                                >         09:30 in Nachricht<br>
                                >       
                                 <CAMvdjurcQc6t=ZfGr=cRL25Xq0J<wbr>e9h9F_TvZXyxVAn3n<br>
                                >         +<a
                                  href="mailto:Dvcgw@mail.gmail.com"
                                  target="_blank" moz-do-not-send="true">Dvcgw@mail.gmail.com</a>>:<br>
                                >         > Hi Ken & Jan,<br>
                                >         ><br>
                                >         > In the cluster we
                                have, there is only one resource
                                running.<br>
                                >         Its a OPT-IN<br>
                                >         > cluster with
                                resource-stickiness set to INFINITY.<br>
                                >         ><br>
                                >         > Just to clarify my
                                question, lets take a scenario where<br>
                                >         there are four<br>
                                >         > nodes N1, N2, N3, N4<br>
                                >         > a. N1 comes up first,
                                starts the cluster.<br>
                                ><br>
                                >         The cluster will start once
                                it has a quorum.<br>
                                ><br>
                                >         > b. N1 Checks that
                                there is no resource running, so it will<br>
                                >         add the<br>
                                >         > resource(R) with the
                                some location constraint(lets say score<br>
                                >         100)<br>
                                >         > c. So Resource(R) runs
                                in N1 now.<br>
                                >         > d. N2 comes up next,
                                checks that resource(R) is already<br>
                                >         running in N1, so<br>
                                >         > it will update the
                                location constraint(lets say score 200)<br>
                                >         > e. N3 comes up next,
                                checks that resource(R) is already<br>
                                >         running in N1, so<br>
                                >         > it will update the
                                location constraint(lets say score 300)<br>
                                ><br>
                                >         See my remark on quorum
                                above.<br>
                                ><br>
                                > Yes you are right, I forgot to
                                mention it.<br>
                                ><br>
                                ><br>
                                >         > f.  N4 comes up next,
                                checks that resource(R) is already<br>
                                >         running in N1, so<br>
                                >         > it will update the
                                location constraint(lets say score 400)<br>
                                >         > g. For the some
                                reason, if N1 goes down, resource(R)
                                shifts<br>
                                >         to N4(as its<br>
                                >         > score is higher than
                                anyone).<br>
                                >         ><br>
                                >         > In this case is it
                                possible to notify the nodes N2, N3 that<br>
                                >         newly elected<br>
                                >         > active node is N4 ?<br>
                                ><br>
                                >         What type of notification,
                                and what would the node do with it?<br>
                                >         Any node in the cluster
                                always has up to date configuration<br>
                                >         information. So it knows
                                the status of the other nodes also.<br>
                                ><br>
                                ><br>
                                > I agree that the node always has
                                upto date configuration information,<br>
                                > but an application or a thread
                                needs to poll for that information. Is<br>
                                > there any way, where the
                                notifications are received through some<br>
                                > action function in RA. ?<br>
                                <br>
                              </div>
                            </div>
                            Ah, I misunderstood your situation, I
                            thought you had a cloned resource.<br>
                            <br>
                            For that, the alerts feature (available in
                            Pacemaker 1.1.15 and later)<br>
                            might be useful:<br>
                            <br>
                            <a
href="http://clusterlabs.org/doc/en-US/Pacemaker/1.1-pcs/html-single/Pacemaker_Explained/index.html#idm139900098676896"
                              rel="noreferrer" target="_blank"
                              moz-do-not-send="true">http://clusterlabs.org/doc/en-<wbr>US/Pacemaker/1.1-pcs/html-sing<wbr>le/Pacemaker_Explained/index.<wbr>html#idm139900098676896</a><br>
                            <div
                              class="m_-4802990479422366187gmail-HOEnZb">
                              <div
                                class="m_-4802990479422366187gmail-h5"><br>
                                <br>
                                ><br>
                                ><br>
                                > Regards,<br>
                                > Sriram.<br>
                                ><br>
                                >         ><br>
                                >         > I went through clone
                                notifications and master-slave, Iooks<br>
                                >         like it either<br>
                                >         > requires identical
                                resources(Anonymous) or Unique or<br>
                                >         Stateful resources to<br>
                                >         > be running<br>
                                >         > in all the nodes of
                                the cluster, where as in our case there<br>
                                >         is only<br>
                                >         > resource running in
                                the whole cluster.<br>
                                ><br>
                                >         Maybe the main reason for
                                not having notifications is that if<br>
                                >         a node fails hard, it won't
                                be able to send out much status<br>
                                >         information to the other
                                nodes.<br>
                                ><br>
                                >         Regards,<br>
                                >         Ulrich<br>
                                ><br>
                                >         ><br>
                                >         > Regards,<br>
                                >         > Sriram.<br>
                                >         ><br>
                                >         ><br>
                                >         ><br>
                                >         ><br>
                                >         > On Mon, Aug 7, 2017 at
                                11:28 AM, Sriram<br>
                                >         <<a
                                  href="mailto:sriram.ec@gmail.com"
                                  target="_blank" moz-do-not-send="true">sriram.ec@gmail.com</a>>
                                wrote:<br>
                                >         ><br>
                                >         >><br>
                                >         >> Thanks Ken, Jan.
                                Will look into the clone notifications.<br>
                                >         >><br>
                                >         >> Regards,<br>
                                >         >> Sriram.<br>
                                >         >><br>
                                >         >> On Sat, Aug 5,
                                2017 at 1:25 AM, Ken Gaillot<br>
                                >         <<a
                                  href="mailto:kgaillot@redhat.com"
                                  target="_blank" moz-do-not-send="true">kgaillot@redhat.com</a>>
                                wrote:<br>
                                >         >><br>
                                >         >>> On Thu,
                                2017-08-03 at 12:31 +0530, Sriram wrote:<br>
                                >         >>> ><br>
                                >         >>> > Hi Team,<br>
                                >         >>> ><br>
                                >         >>> ><br>
                                >         >>> > We have a
                                four node cluster (1 active : 3 standby)
                                in<br>
                                >         our lab for a<br>
                                >         >>> >
                                particular service. If the active node
                                goes down, one of<br>
                                >         the three<br>
                                >         >>> > standby
                                node  becomes active. Now there will be
                                (1<br>
                                >         active :  2<br>
                                >         >>> > standby :
                                1 offline).<br>
                                >         >>> ><br>
                                >         >>> ><br>
                                >         >>> > Is there
                                any way where this newly elected node
                                sends<br>
                                >         notification to<br>
                                >         >>> > the
                                remaining 2 standby nodes about its new
                                status ?<br>
                                >         >>><br>
                                >         >>> Hi Sriram,<br>
                                >         >>><br>
                                >         >>> This depends
                                on how your service is configured in the<br>
                                >         cluster.<br>
                                >         >>><br>
                                >         >>> If you have a
                                clone or master/slave resource, then
                                clone<br>
                                >         notifications<br>
                                >         >>> is probably
                                what you want (not alerts, which is the
                                path<br>
                                >         you were going<br>
                                >         >>> down -- alerts
                                are designed to e.g. email a system<br>
                                >         administrator after<br>
                                >         >>> an important
                                event).<br>
                                >         >>><br>
                                >         >>> For details
                                about clone notifications, see:<br>
                                >         >>><br>
                                >         >>><br>
                                >         <a
                                  href="http://clusterlabs.org/doc/en-US/Pacemaker/1.1-pcs/html-sing"
                                  rel="noreferrer" target="_blank"
                                  moz-do-not-send="true">http://clusterlabs.org/doc/<wbr>en-US/Pacemaker/1.1-pcs/html-<wbr>sing</a><br>
                                >         >>><br>
                                >       
                                 le/Pacemaker_Explained/index.<wbr>html#_clone_resource_agent_req<wbr>uirements<br>
                                >         >>><br>
                                >         >>> The RA must
                                support the "notify" action, which will
                                be<br>
                                >         called when a<br>
                                >         >>> clone instance
                                is started or stopped. See the similar<br>
                                >         section later for<br>
                                >         >>> master/slave
                                resources for additional information.
                                See the<br>
                                >         mysql or<br>
                                >         >>> pgsql resource
                                agents for examples of notify<br>
                                >         implementations.<br>
                                >         >>><br>
                                >         >>> > I was
                                exploring "notification agent" and
                                "notification<br>
                                >         recipient"<br>
                                >         >>> > features,
                                but that doesn't seem to<br>
                                >         work.
                                /etc/sysconfig/notify.sh<br>
                                >         >>> > doesn't
                                get invoked even in the newly elected
                                active<br>
                                >         node.<br>
                                >         >>><br>
                                >         >>> Yep, that's
                                something different altogether -- it's
                                only<br>
                                >         enabled on RHEL<br>
                                >         >>> systems, and
                                solely for backward compatibility with
                                an<br>
                                >         early<br>
                                >         >>> implementation
                                of the alerts interface. The new alerts<br>
                                >         interface is more<br>
                                >         >>> flexible, but
                                it's not designed to send information<br>
                                >         between cluster<br>
                                >         >>> nodes -- it's
                                designed to send information to
                                something<br>
                                >         external to the<br>
                                >         >>> cluster, such
                                as a human, or an SNMP server, or a<br>
                                >         monitoring system.<br>
                                >         >>><br>
                                >         >>><br>
                                >         >>> > Cluster
                                Properties:<br>
                                >         >>> > 
                                cluster-infrastructure: corosync<br>
                                >         >>> > 
                                dc-version: 1.1.17-e2e6cdce80<br>
                                >         >>> > 
                                default-action-timeout: 240<br>
                                >         >>> > 
                                have-watchdog: false<br>
                                >         >>> > 
                                no-quorum-policy: ignore<br>
                                >         >>> > 
                                notification-agent:
                                /etc/sysconfig/notify.sh<br>
                                >         >>> > 
                                notification-recipient:
                                /var/log/notify.log<br>
                                >         >>> > 
                                placement-strategy: balanced<br>
                                >         >>> > 
                                stonith-enabled: false<br>
                                >         >>> > 
                                symmetric-cluster: false<br>
                                >         >>> ><br>
                                >         >>> ><br>
                                >         >>> ><br>
                                >         >>> ><br>
                                >         >>> > I m using
                                the following versions of pacemaker and<br>
                                >         corosync.<br>
                                >         >>> ><br>
                                >         >>> ><br>
                                >         >>> > /usr/sbin
                                # ./pacemakerd --version<br>
                                >         >>> > Pacemaker
                                1.1.17<br>
                                >         >>> > Written
                                by Andrew Beekhof<br>
                                >         >>> > /usr/sbin
                                # ./corosync -v<br>
                                >         >>> > Corosync
                                Cluster Engine, version '2.3.5'<br>
                                >         >>> > Copyright
                                (c) 2006-2009 Red Hat, Inc.<br>
                                >         >>> ><br>
                                >         >>> ><br>
                                >         >>> > Can you
                                please suggest if I m doing anything
                                wrong or if<br>
                                >         there any<br>
                                >         >>> > other
                                mechanisms to achieve this ?<br>
                                >         >>> ><br>
                                >         >>> ><br>
                                >         >>> > Regards,<br>
                                >         >>> > Sriram.<br>
                                <br>
                              </div>
                            </div>
                            <div
                              class="m_-4802990479422366187gmail-HOEnZb">
                              <div
                                class="m_-4802990479422366187gmail-h5">--<br>
                                Ken Gaillot <<a
                                  href="mailto:kgaillot@redhat.com"
                                  target="_blank" moz-do-not-send="true">kgaillot@redhat.com</a>><br>
                                <br>
                                <br>
                                <br>
                                <br>
                                <br>
                                ______________________________<wbr>_________________<br>
                                Users mailing list: <a
                                  href="mailto:Users@clusterlabs.org"
                                  target="_blank" moz-do-not-send="true">Users@clusterlabs.org</a><br>
                                <a
                                  href="http://lists.clusterlabs.org/mailman/listinfo/users"
                                  rel="noreferrer" target="_blank"
                                  moz-do-not-send="true">http://lists.clusterlabs.org/m<wbr>ailman/listinfo/users</a><br>
                                <br>
                                Project Home: <a
                                  href="http://www.clusterlabs.org"
                                  rel="noreferrer" target="_blank"
                                  moz-do-not-send="true">http://www.clusterlabs.org</a><br>
                                Getting started: <a
                                  href="http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf"
                                  rel="noreferrer" target="_blank"
                                  moz-do-not-send="true">http://www.clusterlabs.org/doc<wbr>/Cluster_from_Scratch.pdf</a><br>
                                Bugs: <a
                                  href="http://bugs.clusterlabs.org"
                                  rel="noreferrer" target="_blank"
                                  moz-do-not-send="true">http://bugs.clusterlabs.org</a><br>
                              </div>
                            </div>
                          </blockquote>
                        </div>
                        <br>
                      </div>
                    </div>
                    <br>
                    <fieldset
                      class="m_-4802990479422366187mimeAttachmentHeader"></fieldset>
                    <br>
                    <pre>______________________________<wbr>_________________
Users mailing list: <a class="m_-4802990479422366187moz-txt-link-abbreviated" href="mailto:Users@clusterlabs.org" target="_blank" moz-do-not-send="true">Users@clusterlabs.org</a>
<a class="m_-4802990479422366187moz-txt-link-freetext" href="http://lists.clusterlabs.org/mailman/listinfo/users" target="_blank" moz-do-not-send="true">http://lists.clusterlabs.org/<wbr>mailman/listinfo/users</a>

Project Home: <a class="m_-4802990479422366187moz-txt-link-freetext" href="http://www.clusterlabs.org" target="_blank" moz-do-not-send="true">http://www.clusterlabs.org</a>
Getting started: <a class="m_-4802990479422366187moz-txt-link-freetext" href="http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf" target="_blank" moz-do-not-send="true">http://www.clusterlabs.org/<wbr>doc/Cluster_from_Scratch.pdf</a>
Bugs: <a class="m_-4802990479422366187moz-txt-link-freetext" href="http://bugs.clusterlabs.org" target="_blank" moz-do-not-send="true">http://bugs.clusterlabs.org</a>
</pre>
                  </blockquote>
                  <br>
                </div>
              </div>
            </div>
          </blockquote>
        </div>
        <br>
      </div>
    </blockquote>
    <br>
  </body>
</html>