<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <br>
    Hi Soumya,<br>
    <br>
    I just noticed some of the notes at the bottom.  In particular<br>
    <ul>
      <li>Till glusterfs-3.7, gluster-NFS (gNFS) gets enabled by
        default. The only requirement is that kernel-NFS has to be
        disabled for gluster-NFS to come up. Please disable kernel-NFS
        server and restart glusterd to start gNFS. In case of any issues
        with starting gNFS server, please look at
        /var/log/glusterfs/nfs.log.
      </li>
    </ul>
    If we disable the kernel-NFS on our server and restart glusterd to
    start gNFS will that affect the NFS file system also being served by
    that server (i.e. the single server serves both a glusterFS area and
    an NFS  area)?  Would we also have to disable the kernel-NFS for
    NFS-ganesha?<br>
    <br>
    My second question concerns NFS-ganesha (v 2.3.x) for CentOS 6.8 and
    gluster 3.7.11.  I think I see a couple of possibilities<br>
    <ol>
      <li>I see one possible rpm for version 2.3.3 in
<a class="moz-txt-link-freetext" href="https://mirror.chpc.utah.edu/pub/vault.centos.org/centos/6.8/storage/Source/gluster-3.8/">https://mirror.chpc.utah.edu/pub/vault.centos.org/centos/6.8/storage/Source/gluster-3.8/</a> 
        The other rpm's seem to be for gluster 3.8 packages, so I'm
        wondering if there is a concern for conflict<br>
      </li>
      <li>In one of the links you sent
        (<a class="moz-txt-link-freetext" href="https://buildlogs.centos.org/centos/6/storage/x86_64/gluster-3.7/">https://buildlogs.centos.org/centos/6/storage/x86_64/gluster-3.7/</a>)
        I see an rpm for glusterfs-ganesha-3.7.11 .  Is this a specific
        gluster package for compatibility with ganesha or a ganesha
        package for gluster?<br>
      </li>
    </ol>
    Does either possibility seem more likely to be what I need than the
    other?<br>
    <br>
    Pat<br>
    <br>
    <br>
    <div class="moz-cite-prefix">On 07/07/2017 01:31 PM, Soumya Koduri
      wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:4a73b085-7879-c4a8-944e-a4c8b396e792@redhat.com">Hi,
      <br>
      <br>
      On 07/07/2017 06:16 AM, Pat Haley wrote:
      <br>
      <blockquote type="cite">
        <br>
        Hi All,
        <br>
        <br>
        A follow-up question.  I've been looking at various pages on
        nfs-ganesha
        <br>
        &amp; gluster.  Is there a version of nfs-ganesha that is
        recommended for
        <br>
        use with
        <br>
        <br>
        glusterfs 3.7.11 built on Apr 27 2016 14:09:22
        <br>
        CentOS release 6.8 (Final)
        <br>
      </blockquote>
      <br>
      For glusterfs 3.7, nfs-ganesha-2.3-* version can be used.
      <br>
      <br>
      I see the packages built in centos7 storage sig [1] but not for
      centos6. Request Niels to comment.
      <br>
      <br>
      <br>
      <blockquote type="cite">
        <br>
        Thanks
        <br>
        <br>
        Pat
        <br>
        <br>
        <br>
        On 07/05/2017 11:36 AM, Pat Haley wrote:
        <br>
        <blockquote type="cite">
          <br>
          Hi Soumya,
          <br>
          <br>
          (1) In
          <a class="moz-txt-link-freetext" href="http://mseas.mit.edu/download/phaley/GlusterUsers/TestNFSmount/">http://mseas.mit.edu/download/phaley/GlusterUsers/TestNFSmount/</a>
          <br>
          I've placed the following 2 log files
          <br>
          <br>
          etc-glusterfs-glusterd.vol.log
          <br>
          gdata.log
          <br>
          <br>
          The first has repeated messages about nfs disconnects.  The
          second had
          <br>
          the &lt;fuse_mnt_direcotry&gt;.log name (but not much
          information).
          <br>
          <br>
        </blockquote>
      </blockquote>
      <br>
      Hmm yeah..weird ..there are not much logs in fuse mnt log file.
      <br>
      <br>
      <blockquote type="cite">
        <blockquote type="cite">(2) About the gluster-NFS native
          server:  do you know where we can
          <br>
          find documentation on how to use/install it?  We haven't had
          success
          <br>
          in our searches.
          <br>
          <br>
        </blockquote>
      </blockquote>
      <br>
      Till glusterfs-3.7, gluster-NFS (gNFS) gets enabled by default.
      The only requirement is that kernel-NFS has to be disabled for
      gluster-NFS to come up. Please disable kernel-NFS server and
      restart glusterd to start gNFS. In case of any issues with
      starting gNFS server, please look at /var/log/glusterfs/nfs.log.
      <br>
      <br>
      Thanks,
      <br>
      Soumya
      <br>
      <br>
      <br>
      [1]
      <a class="moz-txt-link-freetext" href="https://buildlogs.centos.org/centos/7/storage/x86_64/gluster-3.7/">https://buildlogs.centos.org/centos/7/storage/x86_64/gluster-3.7/</a>
      <br>
      [2]
      <a class="moz-txt-link-freetext" href="https://buildlogs.centos.org/centos/6/storage/x86_64/gluster-3.7/">https://buildlogs.centos.org/centos/6/storage/x86_64/gluster-3.7/</a>
      <br>
      <br>
      <blockquote type="cite">
        <blockquote type="cite">Thanks
          <br>
          <br>
          Pat
          <br>
          <br>
          <br>
          On 07/04/2017 05:01 AM, Soumya Koduri wrote:
          <br>
          <blockquote type="cite">
            <br>
            <br>
            On 07/03/2017 09:01 PM, Pat Haley wrote:
            <br>
            <blockquote type="cite">
              <br>
              Hi Soumya,
              <br>
              <br>
              When I originally did the tests I ran tcpdump on the
              client.
              <br>
              <br>
              I have rerun the tests, doing tcpdump on the server
              <br>
              <br>
              tcpdump -i any -nnSs 0 host 172.16.1.121 -w
              /root/capture_nfsfail.pcap
              <br>
              <br>
              The results are in the same place
              <br>
              <br>
<a class="moz-txt-link-freetext" href="http://mseas.mit.edu/download/phaley/GlusterUsers/TestNFSmount/">http://mseas.mit.edu/download/phaley/GlusterUsers/TestNFSmount/</a>
              <br>
              <br>
              capture_nfsfail.pcap   has the results from the failed
              touch experiment
              <br>
              capture_nfssucceed.pcap  has the results from the
              successful touch
              <br>
              experiment
              <br>
              <br>
              The brick log files are there too.
              <br>
            </blockquote>
            <br>
            Thanks for sharing. Looks like the error is not generated
            <br>
            @gluster-server side. The permission denied error was caused
            by
            <br>
            either kNFS or by fuse-mnt process or probably by the
            combination.
            <br>
            <br>
            To check fuse-mnt logs, please look at
            <br>
            /var/log/glusterfs/&lt;fuse_mnt_direcotry&gt;.log
            <br>
            <br>
            For eg.: if you have fuse mounted the gluster volume at
            /mnt/fuse-mnt
            <br>
            and exported it via kNFS, the log location for that fuse_mnt
            shall be
            <br>
            at /var/log/glusterfs/mnt-fuse-mnt.log
            <br>
            <br>
            <br>
            Also why not switch to either gluster-NFS native server or
            <br>
            NFS-Ganesha instead of using kNFS, as they are recommended
            NFS
            <br>
            servers to use with gluster?
            <br>
            <br>
            Thanks,
            <br>
            Soumya
            <br>
            <br>
            <blockquote type="cite">
              <br>
              I believe we are using kernel-NFS exporting a fuse mounted
              gluster
              <br>
              volume.  I am having Steve confirm this.  I tried to find
              the fuse-mnt
              <br>
              logs but failed.  Where should I look for them?
              <br>
              <br>
              Thanks
              <br>
              <br>
              Pat
              <br>
              <br>
              <br>
              <br>
              On 07/03/2017 07:58 AM, Soumya Koduri wrote:
              <br>
              <blockquote type="cite">
                <br>
                <br>
                On 06/30/2017 07:56 PM, Pat Haley wrote:
                <br>
                <blockquote type="cite">
                  <br>
                  Hi,
                  <br>
                  <br>
                  I was wondering if there were any additional test we
                  could perform to
                  <br>
                  help debug the group write-permissions issue?
                  <br>
                </blockquote>
                <br>
                Sorry for the delay. Please find response inline --
                <br>
                <br>
                <blockquote type="cite">
                  <br>
                  Thanks
                  <br>
                  <br>
                  Pat
                  <br>
                  <br>
                  <br>
                  On 06/27/2017 12:29 PM, Pat Haley wrote:
                  <br>
                  <blockquote type="cite">
                    <br>
                    Hi Soumya,
                    <br>
                    <br>
                    One example, we have a common working directory
                    dri_fleat in the
                    <br>
                    gluster volume
                    <br>
                    <br>
                    drwxrwsr-x 22 root dri_fleat     4.0K May  1 15:14
                    dri_fleat
                    <br>
                    <br>
                    my user (phaley) does not own that directory but is
                    a member of the
                    <br>
                    group  dri_fleat and should have write permissions.
                    When I go to the
                    <br>
                    nfs-mounted version and try to use the touch command
                    I get the
                    <br>
                    following
                    <br>
                    <br>
                    ibfdr-compute-0-4(dri_fleat)% touch dum
                    <br>
                    touch: cannot touch `dum': Permission denied
                    <br>
                    <br>
                    One of the sub-directories under dri_fleat is "test"
                    which phaley
                    <br>
                    owns
                    <br>
                    <br>
                    drwxrwsr-x  2 phaley   dri_fleat 4.0K May  1 15:16
                    test
                    <br>
                    <br>
                    Under this directory (mounted via nfs) user phaley
                    can write
                    <br>
                    <br>
                    ibfdr-compute-0-4(test)% touch dum
                    <br>
                    ibfdr-compute-0-4(test)%
                    <br>
                    <br>
                    I have put the packet captures in
                    <br>
                    <br>
<a class="moz-txt-link-freetext" href="http://mseas.mit.edu/download/phaley/GlusterUsers/TestNFSmount/">http://mseas.mit.edu/download/phaley/GlusterUsers/TestNFSmount/</a>
                    <br>
                    <br>
                    capture_nfsfail.pcap   has the results from the
                    failed touch
                    <br>
                    experiment
                    <br>
                    capture_nfssucceed.pcap  has the results from the
                    successful touch
                    <br>
                    experiment
                    <br>
                    <br>
                    The command I used for these was
                    <br>
                    <br>
                    tcpdump -i ib0 -nnSs 0 host 172.16.1.119 -w
                    <br>
                    /root/capture_nfstest.pcap
                    <br>
                  </blockquote>
                </blockquote>
                <br>
                I hope these pkts were captured on the node where NFS
                server is
                <br>
                running. Could you please use '-i any' as I do not see
                glusterfs
                <br>
                traffic in the tcpdump.
                <br>
                <br>
                Also looks like NFS v4 is used between client &amp; nfs
                server. Are you
                <br>
                using kernel-NFS here (i.e, kernel-NFS exporting fuse
                mounted gluster
                <br>
                volume)?
                <br>
                If that is the case please capture fuse-mnt logs as
                well. This error
                <br>
                may well be coming from kernel-NFS itself before the
                request is sent
                <br>
                to fuse-mnt process.
                <br>
                <br>
                FWIW, we have below option -
                <br>
                <br>
                Option: server.manage-gids
                <br>
                Default Value: off
                <br>
                Description: Resolve groups on the server-side.
                <br>
                <br>
                I haven't looked into what this option exactly does. But
                it may worth
                <br>
                testing with this option on.
                <br>
                <br>
                Thanks,
                <br>
                Soumya
                <br>
                <br>
                <br>
                <blockquote type="cite">
                  <blockquote type="cite">
                    <br>
                    The brick log files are also in the above link.  If
                    I read them
                    <br>
                    correctly they both funny times.  Specifically I see
                    entries from
                    <br>
                    around 2017-06-27 14:02:37.404865  even though the
                    system time was
                    <br>
                    2017-06-27 12:00:00.
                    <br>
                    <br>
                    One final item, another reply to my post had a link
                    for possible
                    <br>
                    problems that could arise from users belonging to
                    too many group. We
                    <br>
                    have seen the above problem even with a user
                    belonging to only 4
                    <br>
                    groups.
                    <br>
                    <br>
                    Let me know what additional information I can
                    provide.
                    <br>
                  </blockquote>
                </blockquote>
                <br>
                <blockquote type="cite">
                  <blockquote type="cite">
                    <br>
                    Thanks
                    <br>
                    <br>
                    Pat
                    <br>
                    <br>
                    <br>
                    On 06/27/2017 02:45 AM, Soumya Koduri wrote:
                    <br>
                    <blockquote type="cite">
                      <br>
                      <br>
                      On 06/27/2017 10:17 AM, Pranith Kumar Karampuri
                      wrote:
                      <br>
                      <blockquote type="cite">The only problem with
                        using gluster mounted via NFS is that it
                        <br>
                        does not
                        <br>
                        respect the group write permissions which we
                        need.
                        <br>
                        <br>
                        We have an exercise coming up in the a couple of
                        weeks. It seems
                        <br>
                        to me
                        <br>
                        that in order to improve our write times before
                        then, it would be
                        <br>
                        good
                        <br>
                        to solve the group write permissions for gluster
                        mounted via
                        <br>
                        NFS now.
                        <br>
                        We can then revisit gluster mounted via FUSE
                        afterwards.
                        <br>
                        <br>
                        What information would you need to help us force
                        gluster
                        <br>
                        mounted via
                        <br>
                        NFS
                        <br>
                        to respect the group write permissions?
                        <br>
                      </blockquote>
                      <br>
                      Is this owning group or one of the auxiliary
                      groups whose write
                      <br>
                      permissions are not considered? AFAIK, there are
                      no special
                      <br>
                      permission checks done by gNFS server when
                      compared to gluster
                      <br>
                      native
                      <br>
                      client.
                      <br>
                      <br>
                      Could you please provide simple steps to reproduce
                      the issue and
                      <br>
                      collect pkt trace and nfs/brick logs as well.
                      <br>
                      <br>
                      Thanks,
                      <br>
                      Soumya
                      <br>
                    </blockquote>
                    <br>
                  </blockquote>
                  <br>
                </blockquote>
              </blockquote>
              <br>
            </blockquote>
          </blockquote>
          <br>
        </blockquote>
        <br>
      </blockquote>
    </blockquote>
    <br>
    <pre class="moz-signature" cols="72">-- 

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Pat Haley                          Email:  <a class="moz-txt-link-abbreviated" href="mailto:phaley@mit.edu">phaley@mit.edu</a>
Center for Ocean Engineering       Phone:  (617) 253-6824
Dept. of Mechanical Engineering    Fax:    (617) 253-8125
MIT, Room 5-213                    <a class="moz-txt-link-freetext" href="http://web.mit.edu/phaley/www/">http://web.mit.edu/phaley/www/</a>
77 Massachusetts Avenue
Cambridge, MA  02139-4301
</pre>
  </body>
</html>