<html>
  <head>
    <meta content="text/html; charset=windows-1252"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">Hi Bill,<br>
      Can you enable virt-profile setting for your volume and see if
      that helps? You need to enable this optimization when you create
      the volume using ovrit, or use the following command for an
      existing volume:<br>
      <tt><br>
        #gluster volume set &lt;volname&gt; group virt</tt><br>
      <br>
      -Ravi<br>
      <br>
      <br>
      On 02/12/2016 05:22 AM, Bill James wrote:<br>
    </div>
    <blockquote cite="mid:56BD1EB9.5000802@j2.com" type="cite">
      <meta content="text/html; charset=windows-1252"
        http-equiv="Content-Type">
      My apologies, I'm showing how much of a noob I am.<br>
      Ignore last direct to gluster numbers, as that wasn't really
      glusterfs.<br>
      <br>
      <br>
      [root@ovirt2 test ~]# mount -t glusterfs
      ovirt2-ks.test.j2noc.com:/gv1 /mnt/tmp/<br>
      [root@ovirt2 test ~]# time dd if=/dev/zero of=/mnt/tmp/testfile2
      bs=1M count=1000 oflag=direct<br>
      1048576000 bytes (1.0 GB) copied, 65.8596 s, 15.9 MB/s<br>
      <br>
      That's more how I expected, it is pointing to glusterfs
      performance.<br>
      <br>
      <br>
      <br>
      <div class="moz-cite-prefix">On 02/11/2016 03:27 PM, Bill James
        wrote:<br>
      </div>
      <blockquote cite="mid:56BD18D7.8080909@j2.com" type="cite">
        <meta http-equiv="Content-Type" content="text/html;
          charset=windows-1252">
        don't know if it helps, but I ran a few more tests, all from the
        same hardware node.<br>
        <br>
        The VM:<br>
        [root@billjov1 ~]# time dd if=/dev/zero of=/root/testfile bs=1M
        count=1000 oflag=direct<br>
        1048576000 bytes (1.0 GB) copied, 62.5535 s, 16.8 MB/s<br>
        <br>
        Writing directly to gluster volume:<br>
        [root@ovirt2 test ~]# time dd if=/dev/zero
        of=/gluster-store/brick1/gv1/testfile bs=1M count=1000
        oflag=direct<br>
        1048576000 bytes (1.0 GB) copied, 9.92048 s, 106 MB/s<br>
        <br>
        <br>
        Writing to NFS volume:<br>
        [root@ovirt2 test ~]# time dd if=/dev/zero
        of=/mnt/storage/qa/testfile bs=1M count=1000 oflag=direct<br>
        1048576000 bytes (1.0 GB) copied, 10.5776 s, 99.1 MB/s<br>
        <br>
        NFS &amp; Gluster are using the same interface. Tests were not
        run at same time.<br>
        <br>
        This would suggest my problem isn't glusterfs, but the VM
        performance.<br>
        <br>
        <br>
        <br>
        <div class="moz-cite-prefix">On 02/11/2016 03:13 PM, Bill James
          wrote:<br>
        </div>
        <blockquote cite="mid:56BD1589.8090506@j2.com" type="cite">xml
          attached. <br>
          <br>
          <br>
          On 02/11/2016 12:28 PM, Nir Soffer wrote: <br>
          <blockquote type="cite">On Thu, Feb 11, 2016 at 8:27 PM, Bill
            James <a moz-do-not-send="true"
              class="moz-txt-link-rfc2396E"
              href="mailto:bill.james@j2.com">&lt;bill.james@j2.com&gt;</a>
            wrote: <br>
            <blockquote type="cite">thank you for the reply. <br>
              <br>
              We setup gluster using the names associated with  NIC 2
              IP. <br>
                Brick1:
              ovirt1-ks.test.j2noc.com:/gluster-store/brick1/gv1 <br>
                Brick2:
              ovirt2-ks.test.j2noc.com:/gluster-store/brick1/gv1 <br>
                Brick3:
              ovirt3-ks.test.j2noc.com:/gluster-store/brick1/gv1 <br>
              <br>
              That's NIC 2's IP. <br>
              Using 'iftop -i eno2 -L 5 -t' : <br>
              <br>
              dd if=/dev/zero of=/root/testfile bs=1M count=1000
              oflag=direct <br>
              1048576000 bytes (1.0 GB) copied, 68.0714 s, 15.4 MB/s <br>
            </blockquote>
            Can you share the xml of this vm? You can find it in vdsm
            log, <br>
            at the time you start the vm. <br>
            <br>
            Or you can do (on the host): <br>
            <br>
            # virsh <br>
            virsh # list <br>
            (username: vdsm@ovirt password: shibboleth) <br>
            virsh # dumpxml vm-id <br>
            <br>
            <blockquote type="cite">Peak rate
              (sent/received/total):                      281Mb 5.36Mb <br>
              282Mb <br>
              Cumulative (sent/received/total):                   
              1.96GB 14.6MB <br>
              1.97GB <br>
              <br>
              gluster volume info gv1: <br>
                Options Reconfigured: <br>
                performance.write-behind-window-size: 4MB <br>
                performance.readdir-ahead: on <br>
                performance.cache-size: 1GB <br>
                performance.write-behind: off <br>
              <br>
              performance.write-behind: off didn't help. <br>
              Neither did any other changes I've tried. <br>
              <br>
              <br>
              There is no VM traffic on this VM right now except my
              test. <br>
              <br>
              <br>
              <br>
              On 02/10/2016 11:55 PM, Nir Soffer wrote: <br>
              <blockquote type="cite">On Thu, Feb 11, 2016 at 2:42 AM,
                Ravishankar N <a moz-do-not-send="true"
                  class="moz-txt-link-rfc2396E"
                  href="mailto:ravishankar@redhat.com">&lt;ravishankar@redhat.com&gt;</a>
                <br>
                wrote: <br>
                <blockquote type="cite">+gluster-users <br>
                  <br>
                  Does disabling 'performance.write-behind' give a
                  better throughput? <br>
                  <br>
                  <br>
                  <br>
                  On 02/10/2016 11:06 PM, Bill James wrote: <br>
                  <blockquote type="cite">I'm setting up a ovirt cluster
                    using glusterfs and noticing not stellar <br>
                    performance. <br>
                    Maybe my setup could use some adjustments? <br>
                    <br>
                    3 hardware nodes running centos7.2, glusterfs
                    3.7.6.1, ovirt 3.6.2.6-1. <br>
                    Each node has 8 spindles configured in 1 array which
                    is split using LVM <br>
                    with one logical volume for system and one for
                    gluster. <br>
                    They each have 4 NICs, <br>
                       NIC1 = ovirtmgmt <br>
                       NIC2 = gluster  (1GbE) <br>
                  </blockquote>
                </blockquote>
                How do you ensure that gluster trafic is using this nic?
                <br>
                <br>
                <blockquote type="cite">
                  <blockquote type="cite">   NIC3 = VM traffic <br>
                  </blockquote>
                </blockquote>
                How do you ensure that vm trafic is using this nic? <br>
                <br>
                <blockquote type="cite">
                  <blockquote type="cite">I tried with default glusterfs
                    settings <br>
                  </blockquote>
                </blockquote>
                And did you find any difference? <br>
                <br>
                <blockquote type="cite">
                  <blockquote type="cite">and also with: <br>
                    performance.cache-size: 1GB <br>
                    performance.readdir-ahead: on <br>
                    performance.write-behind-window-size: 4MB <br>
                    <br>
                    [root@ovirt3 test scripts]# gluster volume info gv1
                    <br>
                    <br>
                    Volume Name: gv1 <br>
                    Type: Replicate <br>
                    Volume ID: 71afc35b-09d7-4384-ab22-57d032a0f1a2 <br>
                    Status: Started <br>
                    Number of Bricks: 1 x 3 = 3 <br>
                    Transport-type: tcp <br>
                    Bricks: <br>
                    Brick1:
                    ovirt1-ks.test.j2noc.com:/gluster-store/brick1/gv1 <br>
                    Brick2:
                    ovirt2-ks.test.j2noc.com:/gluster-store/brick1/gv1 <br>
                    Brick3:
                    ovirt3-ks.test.j2noc.com:/gluster-store/brick1/gv1 <br>
                    Options Reconfigured: <br>
                    performance.cache-size: 1GB <br>
                    performance.readdir-ahead: on <br>
                    performance.write-behind-window-size: 4MB <br>
                    <br>
                    <br>
                    Using simple dd test on VM in ovirt: <br>
                        dd if=/dev/zero of=/root/testfile bs=1G count=1
                    oflag=direct <br>
                  </blockquote>
                </blockquote>
                block size of 1G?! <br>
                <br>
                Try 1M (our default for storage operations) <br>
                <br>
                <blockquote type="cite">
                  <blockquote type="cite">        1073741824 bytes (1.1
                    GB) copied, 65.9337 s, 16.3 MB/s <br>
                    <br>
                    Another VM not in ovirt using nfs: <br>
                         dd if=/dev/zero of=/root/testfile bs=1G count=1
                    oflag=direct <br>
                            1073741824 bytes (1.1 GB) copied, 27.0079 s,
                    39.8 MB/s <br>
                    <br>
                    <br>
                    Is that expected or is there a better way to set it
                    up to get better <br>
                    performance? <br>
                  </blockquote>
                </blockquote>
                Adding Niels for advice. <br>
                <br>
                <blockquote type="cite">
                  <blockquote type="cite">This email, its contents and
                    .... <br>
                  </blockquote>
                </blockquote>
                Please avoid this, this is a public mailing list,
                everything you write <br>
                here is public. <br>
                <br>
                Nir <br>
              </blockquote>
              I'll have to look into how to remove this sig for this
              mailing list.... <br>
              <br>
              Cloud Services for Business <a moz-do-not-send="true"
                class="moz-txt-link-abbreviated"
                href="http://www.j2.com">www.j2.com</a> <br>
              j2 | eFax | eVoice | FuseMail | Campaigner | KeepItSafe |
              Onebox <br>
              <br>
              <br>
              This email, its contents and attachments contain
              information from j2 Global, <br>
              Inc. and/or its affiliates which may be privileged,
              confidential or <br>
              otherwise protected from disclosure. The information is
              intended to be for <br>
              the addressee(s) only. If you are not an addressee, any
              disclosure, copy, <br>
              distribution, or use of the contents of this message is
              prohibited. If you <br>
              have received this email in error please notify the sender
              by reply e-mail <br>
              and delete the original message and any copies. (c) 2015
              j2 Global, Inc. All <br>
              rights reserved. eFax, eVoice, Campaigner, FuseMail,
              KeepItSafe, and Onebox <br>
              are registered trademarks of j2 Global, Inc. and its
              affiliates. <br>
            </blockquote>
          </blockquote>
          <br>
          <br>
          <fieldset class="mimeAttachmentHeader"></fieldset>
          <br>
          <pre wrap="">_______________________________________________
Users mailing list
<a moz-do-not-send="true" class="moz-txt-link-abbreviated" href="mailto:Users@ovirt.org">Users@ovirt.org</a>
<a moz-do-not-send="true" class="moz-txt-link-freetext" href="http://lists.ovirt.org/mailman/listinfo/users">http://lists.ovirt.org/mailman/listinfo/users</a>
</pre>
        </blockquote>
        <br>
        <p><a moz-do-not-send="true"
href="http://www.j2.com/?utm_source=j2global&amp;utm_medium=xsell-referral&amp;utm_campaign=employeeemail"><span
              style="color:windowtext; text-decoration:none"><img
                moz-do-not-send="true"
src="http://home.j2.com/j2_Global_Cloud_Services/j2_Global_Email_Footer.jpg"
                alt="www.j2.com" height="46" width="391" border="0"></span></a></p>
        <p><span
style="font-size:8.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&quot;;color:gray">This

            email, its contents and attachments contain information from
            <a moz-do-not-send="true"
href="http://www.j2.com/?utm_source=j2global&amp;utm_medium=xsell-referral&amp;utm_campaign=employemail">j2

              Global, Inc</a>. and/or its affiliates which may be
            privileged, confidential or otherwise protected from
            disclosure. The information is intended to be for the
            addressee(s) only. If you are not an addressee, any
            disclosure, copy, distribution, or use of the contents of
            this message is prohibited. If you have received this email
            in error please notify the sender by reply e-mail and delete
            the original message and any copies. © 2015 <a
              moz-do-not-send="true" href="http://www.j2.com/">j2
              Global, Inc</a>. All rights reserved. <a
              moz-do-not-send="true" href="http://www.efax.com/">eFax ®</a>,
            <a moz-do-not-send="true" href="http://www.evoice.com/">eVoice

              ®</a>, <a moz-do-not-send="true"
              href="http://www.campaigner.com/">Campaigner ®</a>, <a
              moz-do-not-send="true" href="http://www.fusemail.com/">FuseMail

              ®</a>, <a moz-do-not-send="true"
              href="http://www.keepitsafe.com/">KeepItSafe ®</a> and <a
              moz-do-not-send="true" href="http://www.onebox.com/">Onebox

              ®</a> are r egistered trademarks of <a
              moz-do-not-send="true" href="http://www.j2.com/">j2
              Global, Inc</a>. and its affiliates.</span></p>
        <br>
        <fieldset class="mimeAttachmentHeader"></fieldset>
        <br>
        <pre wrap="">_______________________________________________
Users mailing list
<a moz-do-not-send="true" class="moz-txt-link-abbreviated" href="mailto:Users@ovirt.org">Users@ovirt.org</a>
<a moz-do-not-send="true" class="moz-txt-link-freetext" href="http://lists.ovirt.org/mailman/listinfo/users">http://lists.ovirt.org/mailman/listinfo/users</a>
</pre>
      </blockquote>
      <br>
      <p><a moz-do-not-send="true"
href="http://www.j2.com/?utm_source=j2global&amp;utm_medium=xsell-referral&amp;utm_campaign=employeeemail"><span
            style="color:windowtext;
            text-decoration:none"><img moz-do-not-send="true"
src="http://home.j2.com/j2_Global_Cloud_Services/j2_Global_Email_Footer.jpg"
              alt="www.j2.com" height="46" width="391" border="0"></span></a></p>
      <p><span
style="font-size:8.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&quot;;color:gray">This
          email, its contents and attachments contain information from <a
            moz-do-not-send="true"
href="http://www.j2.com/?utm_source=j2global&amp;utm_medium=xsell-referral&amp;utm_campaign=employemail">j2
            Global, Inc</a>. and/or its affiliates which may be
          privileged, confidential or otherwise protected from
          disclosure. The information is intended to be for the
          addressee(s) only. If you are not an addressee, any
          disclosure, copy, distribution, or use of the contents of this
          message is prohibited. If you have received this email in
          error please notify the sender by reply e-mail and delete the
          original message and any copies. © 2015 <a
            moz-do-not-send="true" href="http://www.j2.com/">j2 Global,
            Inc</a>. All rights reserved. <a moz-do-not-send="true"
            href="http://www.efax.com/">eFax ®</a>, <a
            moz-do-not-send="true" href="http://www.evoice.com/">eVoice
            ®</a>, <a moz-do-not-send="true"
            href="http://www.campaigner.com/">Campaigner ®</a>, <a
            moz-do-not-send="true" href="http://www.fusemail.com/">FuseMail
            ®</a>, <a moz-do-not-send="true"
            href="http://www.keepitsafe.com/">KeepItSafe ®</a> and <a
            moz-do-not-send="true" href="http://www.onebox.com/">Onebox
            ®</a> are r egistered trademarks of <a
            moz-do-not-send="true" href="http://www.j2.com/">j2 Global,
            Inc</a>. and its affiliates.</span></p>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
Users mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Users@ovirt.org">Users@ovirt.org</a>
<a class="moz-txt-link-freetext" href="http://lists.ovirt.org/mailman/listinfo/users">http://lists.ovirt.org/mailman/listinfo/users</a>
</pre>
    </blockquote>
    <br>
    <br>
  </body>
</html>