<html><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body>
    <p><font face="monospace">Hi Christof,</font></p>
    <p><font face="monospace">thanks! So the preferred way should be not
        to have Filesystem_Id. If I have understood correctly, this is
        anyhow set up by default in CES to 666.666, so the suggested
        procedure during the CES setup should be to manually modify
        gpfs.ganesha.exports.conf and remove this parameter from all the
        exports, is that correct? Is there an easier way, or is there a
        plan to remove the 666.666 default value?</font></p>
    <p><font face="monospace">In our case, we do rely on two separated
        CES clusters (one in prod and one in stand by, so that we can
        perform upgrades with no downtime by migrating the IPs from one
        to the other), so it might be safer to explicilty set
        Filesystem_Id, to ensure consistency among the cluster - would
        that make sense?</font></p>
    <p><font face="monospace">Thanks again!</font></p>
    <p><font face="monospace">Regards</font></p>
    <p><font face="monospace">leo</font><br>
    </p>
    <pre class="moz-signature" cols="72">Paul Scherrer Institut
Dr. Leonardo Sala
Group Leader Data Analysis and Research Infrastructure
Deputy Department Head a.i Science IT Infrastructure and Services department
Science IT Infrastructure and Services department (AWI)
WHGA/036
Forschungstrasse 111
5232 Villigen PSI
Switzerland

Phone: +41 56 310 3369
<a class="moz-txt-link-abbreviated" href="mailto:leonardo.sala@psi.ch">leonardo.sala@psi.ch</a>
<a class="moz-txt-link-abbreviated" href="http://www.psi.ch">www.psi.ch</a></pre>
    <div class="moz-cite-prefix">On 6/28/23 15:22, Christof Schmitt
      wrote:<br>
    </div>
    <blockquote type="cite" cite="mid:46f2c6856498222769ed523483a14c04ffb6a858.camel@us.ibm.com">
      <div>After another discussion it turns out that this parameter is
        not required. While my previous comment is correct, that there
        is the need to have unique handles across file systems, GPFS
        already provides that information and Ganesha handles that
        correctly. So there is no need to set the parameter in the
        Ganesha config.</div>
      <div>Regards,</div>
      <div><br>
      </div>
      <div>Christof</div>
      <div><br>
      </div>
      <div>On Wed, 2023-06-28 at 14:33 +0200, Leonardo Sala wrote:</div>
      <blockquote type="cite">
        <div>
          Hi Christof, thanks a lot! In our case we are exporting
          multiple filesets from 2 filesystems, I guess we should fix
          unique Fileset_IDs for each fileset? What would happen in case
          we just remove the Fileset_id parameter, as suggested by the
        </div>
        <div>
          ZjQcmQRYFpfptBannerStart</div>
        <div dir="ltr" id="pfptBanners6ux0e7" lang="en">
          <div id="pfptBanners6ux0e7">
            <div id="pfptBanners6ux0e7">
              This Message Is From an External Sender </div>
            <div id="pfptBanners6ux0e7">
              This message came from outside your organization. </div>
          </div>
          <div id="pfptBanners6ux0e7">
            <a id="pfptBanners6ux0e7" href="https://us-phishalarm-ewt.proofpoint.com/EWT/v1/PjiDSg!2k-hAt5_AJM6OsYbOMsPTZmPil8BIAHadeloM-6vntuXy4WCtpq_71hHj07-QEB3YfmuMTIe9Ezg-qdLzL_6DvzGmA12g21EJgQ$" moz-do-not-send="true">
              <div class="pfptPrimaryButtons6ux0e7">Report Suspicious </div>
            </a></div>
          <div>
             </div>
        </div>
        <div>
          ZjQcmQRYFpfptBannerEnd</div>
        <p>Hi Christof,</p>
        <p>thanks a lot! In our case we are exporting multiple filesets
          from 2 filesystems, I guess we should fix unique Fileset_IDs
          for each fileset? What would happen in case we just remove the
          Fileset_id parameter, as suggested by the ganesha docs?</p>
        <p>Regards</p>
        <p>leo<br>
        </p>
        <pre>Paul Scherrer Institut</pre>
        <pre>Dr. Leonardo Sala</pre>
        <pre>Group Leader Data Analysis and Research Infrastructure</pre>
        <pre>Deputy Department Head a.i Science IT Infrastructure and Services department</pre>
        <pre>Science IT Infrastructure and Services department (AWI)</pre>
        <pre>WHGA/036</pre>
        <pre>Forschungstrasse 111</pre>
        <pre>5232 Villigen PSI</pre>
        <pre>Switzerland</pre>
        <pre>
</pre>
        <pre>Phone: +41 56 310 3369</pre>
        <a class="moz-txt-link-abbreviated" href="mailto:leonardo.sala@psi.ch" moz-do-not-send="true">
          <pre>leonardo.sala@psi.ch</pre>
        </a>
        <pre>
</pre>
        <a class="moz-txt-link-abbreviated" href="http://www.psi.ch" moz-do-not-send="true">
          <pre>www.psi.ch</pre>
        </a>
        <div class="moz-cite-prefix">On 6/28/23 14:22, Christof Schmitt
          wrote:<br>
        </div>
        <blockquote type="cite" cite="mid:558222eeef7cd5169214368c9a04af813509b086.camel@us.ibm.com">
          <div>The "FileSystem_Id" is a unique identifier for the file
            system. The technical background is that Ganesha asks the
            file system for a file handle, but that is only unique
            within the file system. If there are NFS exports on
            different file systems, there needs to be a way to make the
            file handles unique across multiple file systems. So if
            there are NFS exports on different file systems, this
            parameter should be set with a unique value for each file
            system. If there is only one file system with NFS exports,
            then this should not be necessary.</div>
          <div><br>
          </div>
          <div>Regards,</div>
          <div><br>
          </div>
          <div>Christof</div>
          <div><br>
          </div>
          <div>On Wed, 2023-06-28 at 08:53 +0200, Leonardo Sala wrote:</div>
          <blockquote type="cite">
            <div>Hi Ed, thanks! In our case we do have unique export
              ids, but the same fsid, and this seems to create issues.
              Also, reading Ganesha docs, I can see [*]: FileSystem_ID
              EXPORT Option There is an EXPORT config option,
              FileSystem_ID. This really
            </div>
            <div>ZjQcmQRYFpfptBannerStart</div>
            <div dir="ltr" id="pfptBannerls5jz08" lang="en">
              <div id="pfptBannerls5jz08">
                <div id="pfptBannerls5jz08">This Message Is From an
                  External Sender </div>
                <div id="pfptBannerls5jz08">This message came from
                  outside your organization. </div>
              </div>
              <div id="pfptBannerls5jz08"><a id="pfptBannerls5jz08" href="https://us-phishalarm-ewt.proofpoint.com/EWT/v1/PjiDSg!2k-ror9_yrma2gYRGMtvzfBw-f7-KrLp6ZZNwHChWJfOvNhVe0t6WfmouxHnuoc3gtMvXaxWV3AbFrgYAQT1JIEKoq1hoJ8cr1c$" moz-do-not-send="true">
                  <div class="pfptPrimaryButtonls5jz08">Report Suspicious
                  </div>
                </a></div>
              <div> </div>
            </div>
            <div>ZjQcmQRYFpfptBannerEnd</div>
            <p>Hi Ed,</p>
            <p>thanks! In our case we do have unique export ids, but the
              same fsid, and this seems to create issues. Also, reading
              Ganesha docs, I can see [*]:</p>
            <h3><span>FileSystem_ID EXPORT Option</span></h3>
            <p>There is an EXPORT config option, FileSystem_ID. This
              really should not be used, all it does it designate an
              fsid to be used with the attributes of all objects in the
              export. It will be folded to fit into NFSv3. Because it
              applies to the entire export, it prevents exporting
              multiple file systems since there will likely be issues
              with collision of inode numbers on the client.</p>
            <p>so before touching the defaults in GPFS CES configuration
              I would like some guidance or experiences from this mlist
              :)</p>
            <p>cheers</p>
            <p>leo<br>
            </p>
            <p>[*] <a class="moz-txt-link-freetext" href="https://github.com/nfs-ganesha/nfs-ganesha/wiki/File-Systems#FileSystem_ID_EXPORT_Option" moz-do-not-send="true">https://github.com/nfs-ganesha/nfs-ganesha/wiki/File-Systems#FileSystem_ID_EXPORT_Option</a><br>
            </p>
            <p><br>
            </p>
            <pre>Paul Scherrer Institut</pre>
            <pre>Dr. Leonardo Sala</pre>
            <pre>Group Leader Data Analysis and Research Infrastructure</pre>
            <pre>Deputy Department Head a.i Science IT Infrastructure and Services department</pre>
            <pre>Science IT Infrastructure and Services department (AWI)</pre>
            <pre>WHGA/036</pre>
            <pre>Forschungstrasse 111</pre>
            <pre>5232 Villigen PSI</pre>
            <pre>Switzerland</pre>
            <pre>Phone: +41 56 310 3369</pre>
            <a class="moz-txt-link-abbreviated" href="mailto:leonardo.sala@psi.ch" moz-do-not-send="true">
              <pre>leonardo.sala@psi.ch</pre>
            </a><a class="moz-txt-link-abbreviated" href="http://www.psi.ch" moz-do-not-send="true">
              <pre>www.psi.ch</pre>
            </a>
            <div class="moz-cite-prefix">On 6/27/23 19:41, Wahl, Edward
              wrote:<br>
            </div>
            <blockquote type="cite" cite="mid:CH0PR01MB6825595F142F5D9C9F269324A827A@CH0PR01MB6825.prod.exchangelabs.com">
              <meta name="Generator" content="Microsoft Word 15
                (filtered medium)">
              <div class="WordSection1">
                <p class="MsoNormal"><span>I vaguely recall seeing this
                    and testing it.  My notes to myself say: ‘As long as
                    the export_id is unique, you are fine.’   See the
                    manuals, ganesha loves Camel Case so it’s more than
                    likely actually “Export_Id” or some such.
                  </span></p>
                <p class="MsoNormal"><span> </span></p>
                <p class="MsoNormal"><span>Ed Wahl</span></p>
                <p class="MsoNormal"><span>Ohio Supercomputer Center</span></p>
                <p class="MsoNormal"><span> </span></p>
                <div>
                  <p class="MsoNormal"><b><span>From:</span></b><span>
                      gpfsug-discuss <a class="moz-txt-link-rfc2396E" href="mailto:gpfsug-discuss-bounces@gpfsug.org" moz-do-not-send="true"><gpfsug-discuss-bounces@gpfsug.org></a>
                      <b>On Behalf Of </b>Leonardo Sala<br>
                      <b>Sent:</b> Tuesday, June 27, 2023 10:18 AM<br>
                      <b>To:</b> <a class="moz-txt-link-abbreviated
                        moz-txt-link-freetext" href="mailto:gpfsug-discuss@spectrumscale.org" moz-do-not-send="true">gpfsug-discuss@spectrumscale.org</a><br>
                      <b>Subject:</b> [gpfsug-discuss] CES, Ganesha, and
                      Filesystem_id</span></p>
                </div>
                <p class="MsoNormal"> </p>
                <div>
                  <p class="MsoNormal"><span>Hallo, we are checking our
                      current CES configuration, and we noticed that by
                      default GPFS puts always Filesystem_Id=666. 666
                      [*], no matter which Export_Id value the export
                      has. To my understanding (which is poor!), this
                      means that all clients </span></p>
                </div>
                <div>
                  <p class="MsoNormal"><span></span></p>
                </div>
                <p><span>Hallo,</span></p>
                <p><span>we are checking our current CES configuration,
                    and we noticed that by default GPFS puts always
                    Filesystem_Id=666.666 [*], no matter which Export_Id
                    value the export has. To my understanding (which is
                    poor!), this means that all clients will see all our
                    exports (~20) with the same device number, creating
                    various possible issues (e.g. file state handles).
                    Questions:</span></p>
                <p><span>* is there a reason for such default value? If
                    we change it, are there unpleasant effects we could
                    see?</span></p>
                <p><span>* what would be a reasonable value? Looking
                    around I saw that Filesystem_Id =
                    Export_Id.Export_Id is quite common, with the
                    possible issue of using the forbidden 152.152 [**]</span></p>
                <p><span>* what happens if we actually remove the
                    Filesytem_Id parameter from
                    gpfs.ganesha.exports.conf?</span></p>
                <p><span>* is there a way to modify Filesystem_Id in
                    gpfs.ganesha.exports.conf without editing the file,
                    eg using mmnfs commands (seems not, but I might be
                    mistaken)?</span></p>
                <p><span>Thanks a lot!</span></p>
                <p><span>cheers</span></p>
                <p><span>leo</span></p>
                <p> </p>
                <p><span>[*] <a href="https://www.ibm.com/docs/en/storage-scale/5.0.4?topic=exports-making-bulk-changes-nfs" class="moz-txt-link-freetext" moz-do-not-send="true">https://www.ibm.com/docs/en/storage-scale/5.0.4?topic=exports-making-bulk-changes-nfs</a></span></p>
                <p><span>[**] <a href="https://github.com/nfs-ganesha/nfs-ganesha/issues/615" class="moz-txt-link-freetext" moz-do-not-send="true">https://github.com/nfs-ganesha/nfs-ganesha/issues/615</a></span></p>
                <pre>-- </pre>
                <pre>Paul Scherrer Institut</pre>
                <pre>Dr. Leonardo Sala</pre>
                <pre>Group Leader Data Analysis and Research Infrastructure</pre>
                <pre>Deputy Department Head a.i Science IT Infrastructure and Services department</pre>
                <pre>Science IT Infrastructure and Services department (AWI)</pre>
                <pre>WHGA/036</pre>
                <pre>Forschungstrasse 111</pre>
                <pre>5232 Villigen PSI</pre>
                <pre>Switzerland</pre>
                <pre> </pre>
                <pre>Phone: +41 56 310 3369</pre>
                <a href="mailto:leonardo.sala@psi.ch" class="moz-txt-link-freetext" moz-do-not-send="true">
                  <pre>leonardo.sala@psi.ch</pre>
                </a><a href="http://www.psi.ch" moz-do-not-send="true">
                  <pre>www.psi.ch</pre>
                </a></div>
              <br>
              <fieldset class="moz-mime-attachment-header"></fieldset>
              <pre>_______________________________________________</pre>
              <pre>gpfsug-discuss mailing list</pre>
              <pre>gpfsug-discuss at gpfsug.org</pre>
              <a class="moz-txt-link-freetext" href="http://gpfsug.org/mailman/listinfo/gpfsug-discuss_gpfsug.org" moz-do-not-send="true">
                <pre>http://gpfsug.org/mailman/listinfo/gpfsug-discuss_gpfsug.org</pre>
              </a></blockquote>
            <pre>_______________________________________________</pre>
            <pre>gpfsug-discuss mailing list</pre>
            <pre>gpfsug-discuss at gpfsug.org</pre>
            <a href="http://gpfsug.org/mailman/listinfo/gpfsug-discuss_gpfsug.org" moz-do-not-send="true">
              <pre>http://gpfsug.org/mailman/listinfo/gpfsug-discuss_gpfsug.org</pre>
            </a>
            <pre> </pre>
          </blockquote>
          <br>
          <fieldset class="moz-mime-attachment-header"></fieldset>
          <pre>_______________________________________________</pre>
          <pre>gpfsug-discuss mailing list</pre>
          <pre>gpfsug-discuss at gpfsug.org</pre>
          <a class="moz-txt-link-freetext" href="http://gpfsug.org/mailman/listinfo/gpfsug-discuss_gpfsug.org" moz-do-not-send="true">
            <pre>http://gpfsug.org/mailman/listinfo/gpfsug-discuss_gpfsug.org</pre>
          </a>
          <pre>
</pre>
        </blockquote>
      </blockquote>
    </blockquote>
  </body>
</html>