<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
  <head>
    <meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
    <title></title>
  </head>
  <body bgcolor="#ffffff" text="#000000">
    lol... How many large databases have you managed?  Once evicted,
    MySQL will be restarted by Pacemaker so all the caches will be
    cold.  On a 128GB db server with an InnoDB buffer pool of >100
    GB, warming the cache takes hours and during that time the db is
    much slower.  Since the slave has been evicted because it was
    lagging, restarting it will not help at all.<br>
    <br>
    Regards,<br>
    <br>
    Yves<br>
    <br>
    <br>
    <br>
    On 11-11-12 12:24 PM, Michael Marrotte wrote:
    <blockquote
      cite="mid:746B5B07-9E24-40AC-B3FE-791DC5F18FC3@gmail.com"
      type="cite">
      <div>Why's evicted terrible?  It optional and simply removes the
        slave node from the cluster if it's lagging too far behind the
        master.  The collocation IP can float to another slave in the
        cluster or go offline.  I thought that's what you're looking
        for.  I can post a config if you're interested.</div>
      <div><br>
        Sent from my iPhone</div>
      <div><br>
        On Nov 12, 2011, at 11:51 AM, Yves Trudeau <<a
          moz-do-not-send="true" href="mailto:y.trudeau@videotron.ca">y.trudeau@videotron.ca</a>>
        wrote:<br>
        <br>
      </div>
      <blockquote type="cite">
        <div>
          <meta content="text/html; charset=UTF-8"
            http-equiv="Content-Type">
          Hi, <br>
             Setting evict_outdated_slaves is terrible but then how do
          you remove the read VIPs from slaves that are outdated.  If
          there is a way to do it, tell me :)  I am always open to do
          things in a simpler way.  This solution works without evicting
          outdated slaves.   The current mysql RA does behave correctly
          HA wise but it is less than optimal database wise.  I mean in
          small installations it is probably ok, but at Percona, I hit
          many high end installations where the current behavior is not
          adequate.  I consider myself more as a database specialist
          than an HA one so I may be more sensitive to these problems.  
          <br>
          <br>
          Regards,   <br>
          <br>
          Yves<br>
          <br>
          On 11-11-12 09:19 AM, Michael Marrotte wrote:
          <blockquote
cite="mid:CA+EJDnA2ePgo9PQ3OT0tUmzZS7-mTKuvr7Ab8m57DrRjrWABmA@mail.gmail.com"
            type="cite">Are you using <span class="s">evict_outdated_slaves</span>?<br>
            <pre><div style="background-color: transparent;" class="line" id="LC296"><span class="s"><parameter name="evict_outdated_slaves" unique="0" required="0"></span></div>
<div style="background-color: transparent;" class="line" id="LC297"><span class="s"><longdesc lang="en"></span></div><div style="background-color: transparent;" class="line" id="LC298"><span class="s">If set to true, any slave which is more than max_slave_lag seconds</span></div>
<div style="background-color: transparent;" class="line" id="LC299"><span class="s">behind the master has its MySQL instance shut down. If this parameter</span></div><div style="background-color: transparent;" class="line" id="LC300">
<span class="s">is set to false in a primitive or clone resource, it is simply</span></div><div style="background-color: transparent;" class="line" id="LC301"><span class="s">ignored. If set to false in a master/slave resource, then exceeding</span></div>
<div style="background-color: transparent;" class="line" id="LC302"><span class="s">the maximum slave lag will merely push down the master preference so</span></div><div style="background-color: transparent;" class="line" id="LC303">
<span class="s">the lagging slave is never promoted to the new master.</span></div><div style="background-color: transparent;" class="line" id="LC304"><span class="s"></longdesc></span></div><div style="background-color: transparent;" class="line" id="LC305">
<span class="s"><shortdesc lang="en">Determines whether to shut down badly lagging</span></div><div style="background-color: transparent;" class="line" id="LC306"><span class="s">slaves</shortdesc></span></div>
<div style="background-color: transparent;" class="line" id="LC307"><span class="s"><content type="boolean" default="${OCF_RESKEY_evict_outdated_slaves_default}" /></span></div><div style="background-color: transparent;" class="line" id="LC308">
<span class="s"></parameter></span></div></pre>
            <br>
            <br>
            <div class="gmail_quote">On Fri, Nov 11, 2011 at 9:57 PM,
              Yves Trudeau <span dir="ltr"><<a
                  moz-do-not-send="true"
                  href="mailto:y.trudeau@videotron.ca">y.trudeau@videotron.ca</a>></span>
              wrote:<br>
              <blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt
                0.8ex; border-left: 1px solid rgb(204, 204, 204);
                padding-left: 1ex;">
                <div bgcolor="#ffffff" text="#000000"> Because that's
                  not enough, if a slave lags behind too much, I want to
                  remove the vip but not stop the slave.  <br>
                  <div>
                    <div class="h5"> <br>
                      On 11-11-11 06:12 PM, Michael Marrotte wrote:
                      <blockquote type="cite">Why don't you simply
                        colocate VIP's with the master/slave roles?<br>
                        <br>
                        <div class="gmail_quote">On Fri, Nov 11, 2011 at
                          5:18 PM, Yves Trudeau <span dir="ltr"><<a
                              moz-do-not-send="true"
                              href="mailto:y.trudeau@videotron.ca"
                              target="_blank">y.trudeau@videotron.ca</a>></span>
                          wrote:<br>
                          <blockquote class="gmail_quote" style="margin:
                            0pt 0pt 0pt 0.8ex; border-left: 1px solid
                            rgb(204, 204, 204); padding-left: 1ex;">Hi,<br>
                              I created a fork of the resource-agents
                            and modified the mysql RA to supports VIP.
                             The support for read/write VIPs is
                            something that currently limits the use of
                            pacemaker to manage mysql replication.  I
                            ran some basic tests on the modified agent
                            and it seems behaving sanely so far but more
                            tests will be needed (next week).  I also
                            modified a few things in the RA that were
                            not optimal in my view.   Here's the URL for
                            my fork:<br>
                            <br>
                            git://<a moz-do-not-send="true"
                              href="http://github.com/y-trudeau/resource-agents.git"
                              target="_blank">github.com/y-trudeau/resource-agents.git</a><br>
                            <a moz-do-not-send="true"
                              href="https://y-trudeau@github.com/y-trudeau/resource-agents.git"
                              target="_blank">https://y-trudeau@github.com/y-trudeau/resource-agents.git</a><br>
                            <br>
                            In order to use it, you need to adapt the
                            pacemaker configuration, here's the one I am
                            using:<br>
                            <br>
                            <a moz-do-not-send="true"
                              href="http://pastebin.com/QPJPWDhx"
                              target="_blank">http://pastebin.com/QPJPWDhx</a><br>
                            <br>
                            Next, apart from running more tests,  I want
                            to do the following:<br>
                            <br>
                            - review user privileges used, need to lower
                            the ones of replication_user<br>
                            - add support for MHA  (<a
                              moz-do-not-send="true"
                              href="http://code.google.com/p/mysql-master-ha/"
                              target="_blank">http://code.google.com/p/mysql-master-ha/</a>)<br>
                            - run a beautifier for more uniformity in
                            style<br>
                            <br>
                            Reviews and comments are of course welcome.<br>
                            <br>
                            Regards,<br>
                            <br>
                            Yves<br>
                            <br>
                            <br>
                            <br>
                            <br>
_______________________________________________<br>
                            Pacemaker mailing list: <a
                              moz-do-not-send="true"
                              href="mailto:Pacemaker@oss.clusterlabs.org"
                              target="_blank">Pacemaker@oss.clusterlabs.org</a><br>
                            <a moz-do-not-send="true"
                              href="http://oss.clusterlabs.org/mailman/listinfo/pacemaker"
                              target="_blank">http://oss.clusterlabs.org/mailman/listinfo/pacemaker</a><br>
                            <br>
                            Project Home: <a moz-do-not-send="true"
                              href="http://www.clusterlabs.org"
                              target="_blank">http://www.clusterlabs.org</a><br>
                            Getting started: <a moz-do-not-send="true"
href="http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf"
                              target="_blank">http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf</a><br>
                            Bugs: <a moz-do-not-send="true"
href="http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker"
                              target="_blank">http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker</a><br>
                          </blockquote>
                        </div>
                        <br>
                        <pre><fieldset></fieldset>
_______________________________________________
Pacemaker mailing list: <a moz-do-not-send="true" href="mailto:Pacemaker@oss.clusterlabs.org" target="_blank">Pacemaker@oss.clusterlabs.org</a>
<a moz-do-not-send="true" href="http://oss.clusterlabs.org/mailman/listinfo/pacemaker" target="_blank">http://oss.clusterlabs.org/mailman/listinfo/pacemaker</a>

Project Home: <a moz-do-not-send="true" href="http://www.clusterlabs.org" target="_blank">http://www.clusterlabs.org</a>
Getting started: <a moz-do-not-send="true" href="http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf" target="_blank">http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf</a>
Bugs: <a moz-do-not-send="true" href="http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker" target="_blank">http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker</a>
</pre>
                      </blockquote>
                      <br>
                    </div>
                  </div>
                </div>
                <br>
                _______________________________________________<br>
                Pacemaker mailing list: <a moz-do-not-send="true"
                  href="mailto:Pacemaker@oss.clusterlabs.org">Pacemaker@oss.clusterlabs.org</a><br>
                <a moz-do-not-send="true"
                  href="http://oss.clusterlabs.org/mailman/listinfo/pacemaker"
                  target="_blank">http://oss.clusterlabs.org/mailman/listinfo/pacemaker</a><br>
                <br>
                Project Home: <a moz-do-not-send="true"
                  href="http://www.clusterlabs.org" target="_blank">http://www.clusterlabs.org</a><br>
                Getting started: <a moz-do-not-send="true"
                  href="http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf"
                  target="_blank">http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf</a><br>
                Bugs: <a moz-do-not-send="true"
href="http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker"
                  target="_blank">http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker</a><br>
                <br>
              </blockquote>
            </div>
            <br>
            <pre wrap=""><fieldset class="mimeAttachmentHeader"></fieldset>
_______________________________________________
Pacemaker mailing list: <a moz-do-not-send="true" class="moz-txt-link-abbreviated" href="mailto:Pacemaker@oss.clusterlabs.org">Pacemaker@oss.clusterlabs.org</a>
<a moz-do-not-send="true" class="moz-txt-link-freetext" href="http://oss.clusterlabs.org/mailman/listinfo/pacemaker">http://oss.clusterlabs.org/mailman/listinfo/pacemaker</a>

Project Home: <a moz-do-not-send="true" class="moz-txt-link-freetext" href="http://www.clusterlabs.org">http://www.clusterlabs.org</a>
Getting started: <a moz-do-not-send="true" class="moz-txt-link-freetext" href="http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf">http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf</a>
Bugs: <a moz-do-not-send="true" class="moz-txt-link-freetext" href="http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker">http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker</a>
</pre>
          </blockquote>
          <br>
        </div>
      </blockquote>
      <blockquote type="cite">
        <div><span>_______________________________________________</span><br>
          <span>Pacemaker mailing list: <a moz-do-not-send="true"
              href="mailto:Pacemaker@oss.clusterlabs.org">Pacemaker@oss.clusterlabs.org</a></span><br>
          <span><a moz-do-not-send="true"
              href="http://oss.clusterlabs.org/mailman/listinfo/pacemaker">http://oss.clusterlabs.org/mailman/listinfo/pacemaker</a></span><br>
          <span></span><br>
          <span>Project Home: <a moz-do-not-send="true"
              href="http://www.clusterlabs.org">http://www.clusterlabs.org</a></span><br>
          <span>Getting started: <a moz-do-not-send="true"
              href="http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf">http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf</a></span><br>
          <span>Bugs: <a moz-do-not-send="true"
href="http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker">http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker</a></span><br>
        </div>
      </blockquote>
      <pre wrap="">
<fieldset class="mimeAttachmentHeader"></fieldset>
_______________________________________________
Pacemaker mailing list: <a class="moz-txt-link-abbreviated" href="mailto:Pacemaker@oss.clusterlabs.org">Pacemaker@oss.clusterlabs.org</a>
<a class="moz-txt-link-freetext" href="http://oss.clusterlabs.org/mailman/listinfo/pacemaker">http://oss.clusterlabs.org/mailman/listinfo/pacemaker</a>

Project Home: <a class="moz-txt-link-freetext" href="http://www.clusterlabs.org">http://www.clusterlabs.org</a>
Getting started: <a class="moz-txt-link-freetext" href="http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf">http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf</a>
Bugs: <a class="moz-txt-link-freetext" href="http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker">http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker</a>
</pre>
    </blockquote>
    <br>
  </body>
</html>