<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Dear Darrell,</p>
    <p>Thanks again for the suggestions. We used some of the options
      suggested for our case and in the meantime we also set the
      round-robin configuration in order to have a single pointer in the
      samba share configuration that then picks one of the bricks
      (single host name in our DNS each brick).</p>
    <p>This has improved the situation and the bricks are developing
      less entries. But less is not really "none" as unfortunately we
      have a behaviour observed:</p>
    <p>We very frequently have directories not healing and then the
      files and dirs under this "base path".</p>
    <p>Sometimes one brick reports "New Folder" in the path, and the
      others have already a name for it.</p>
    <p>So we see a pattern also with renamed directories. I think that
      we have an issue with renaming taking place slowly on one brick
      (and it is not necessary the same one) while the others have
      already the correct name.</p>
    <p>Rebooting solves this, or renaming the file on the volume (or in
      case of new folder create a "New Folder" in the path the brick
      which is wrong expects it), issue listing, (delete if "New
      Folder") rename the folder with old name on the "wrong" brick,
      listing, re-renaming and the entries are healed. At this point the
      names match and everything is back. What is not fully
      understandable for me (because I surely miss something
      fundamental) is that the gfid should be the same everywhere and I
      would not expect that a rename changes that. So what is seen as
      mismatching is just the "name attribute", once this matches,
      everything is back fine.</p>
    <p>Sorry for the simplistic exposure of facts, but I hope you have
      some suggestions anyway. Perhaps it would be better if I send a
      post related to renamings. I saw folks setting open-behind to off,
      tried that but no results. Could be also that we mount incorrectly
      on our ctdbs? Again I apologize as I understand that the points of
      failure can be many, but we are trying to do something with
      data/logs and observations that we have available.</p>
    <p>Best regards<br>
      Ilias<br>
    </p>
    <div class="moz-cite-prefix">Am 10.04.24 um 18:40 schrieb Darrell
      Budic:<br>
    </div>
    <blockquote type="cite"
      cite="mid:38664E17-2F70-4DA0-8231-DF2D71092B29@onholyground.com">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      I would strongly recommend running the glusterfs servers directly
      on bare metal instead of in VMs. Check out Ovirt, especially its
      hybrid cluster model. While it’s not currently well maintained, it
      works fine on your class of hardware and fully supports this model
      of gluster on the bare metal and VMs running on the same hosts.
      And we may see it get some more support after the VMWare buyout,
      who knows?<br id="lineBreakAtBeginningOfMessage">
      <div><br>
      </div>
      <div>Gluster isn’t known for small file performance, but hunt
        through the archives for specific tuning hints. And if you’re
        using it to host the VM image files, you’re making that problem
        because the files shared by gluster are large. More cache and
        write (behind) buffers can help, and 10G or better networking
        would be something you want to do if you can afford it. Going to
        2x1G LAGs can help a tiny bit, but you really want the lower
        latency from a faster physical media if you can get it.</div>
      <div><br>
      </div>
      <div>If you are not already using tuned to set virtual-guest
        profiles on your VMs (and virtual-host on the hosts), I’d look
        into that as well. Set the disk elevator to ’none’ on the VMs as
        well.</div>
      <div><br>
      </div>
      <div><br>
        <blockquote type="cite">
          <div>On Apr 10, 2024, at 10:07 AM, Ilias Chasapakis forumZFD
            <a class="moz-txt-link-rfc2396E" href="mailto:chasapakis@forumZFD.de"><chasapakis@forumZFD.de></a> wrote:</div>
          <br class="Apple-interchange-newline">
          <div>
            <meta http-equiv="Content-Type"
              content="text/html; charset=UTF-8">
            <div>
              <p>Dear Darrell,</p>
              <p>Dear ...,<br>
                <br>
                Many thanks for the prompt reply. Here some of the
                additional information requested (please feel free to
                ask for more if needed)<br>
                <br>
              </p>
              <blockquote type="cite">CPU info:<br>
                Hosts<br>
                1. Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz (20 cores)
                hw RAID 1 (adaptec)<br>
                2. Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz (20 cores)
                hw RAID 1 (adaptec)<br>
                3. Intel(R) Xeon(R) Silver 4112 CPU @ 2.60GHz (8 cores)
                hw RAID 1 (adaptec)<br>
                <br>
                GlusterFS VMs<br>
                1. 4 cores,  10 GB RAM, CPU model Broadwell<br>
                2. 4 cores,  10 GB RAM, CPU model Broadwell<br>
                3. 4 cores,  10 GB RAM, CPU model: host passthrough
                value<br>
                <br>
                Network info<br>
                Physical connection between gluster nodes in a heartbeat
                network that comprises a new Cisco switch.<br>
                TCP connections with 1Gbit links<br>
                Virtual default connectivity with virtio drivers for the
                NICs and Macvtap connection to use the host´s
                connectivity.<br>
                No errors or lost packages are recorded between the VMs
                or the hosts. Quick iperf tests (between glusters and
                ctdbs-glusters show now evident issues).<br>
              </blockquote>
              <blockquote type="cite"> <br>
                Workload:<br>
                An instantaneous from this moment which can be
                considered a peak time is around 450 files open on the
                volume.<br>
                In terms of "litteral" load we notice cpu peaks mostly
                related to the shd process<br>
              </blockquote>
              <br>
              All disks use virtIO drivers (virtIO disks).<br>
              <br>
              The file system on all nodes is XFS (not ZFS)<br>
              <br>
              Other than the clients on the gluster nodes themselves
              there are clients on ctdbs that mount the gluster volume
              and then expose it via smb to Windows clients (user
              profiles included for roaming profiles).<br>
              ctdbs reach the glusters through the heartbeat network<br>
              <br>
              We are considering to move the glusters to a network with
              existing DNS capabilities in order to create a round-robin
              configuration by assigning hosts by IP to a single
              hostname to use it then for the mounts configuration of
              the ctdbs.<br>
              The reasoning/hope behind that we would minimize access
              time and sync issues.<br>
              <br>
              Thank you for the information about the "sharding" we will
              take this into account and consider pros and cons in the
              current situation, epsecially because turning back is not
              easy afterwards. Also our main problem is mainly not with
              big files, but with a large quantity of small files.<br>
              <br>
              We could gladly make use of some of the options you
              suggested after we assess the situation again. We welcome
              any further suggestion in the meantime.
              <div><br class="webkit-block-placeholder">
              </div>
              <p>Ilias<br>
              </p>
              <div class="moz-cite-prefix">Am 09.04.24 um 18:26 schrieb
                Darrell Budic:<br>
              </div>
              <blockquote type="cite"
cite="mid:FDFC289F-5232-4D84-8CDB-BA10A3BAB2AB@onholyground.com">
                <meta http-equiv="content-type"
                  content="text/html; charset=UTF-8">
                The big one I see of you is to investigate and enable
                sharding. It can improve performance and makes it much
                easier to heal VM style workloads. Be aware that once
                you turn it on, you can’t go back easily, and you need
                to copy the VM disk images around to get them to be
                sharded before it will show any real effect. A couple
                other recommendations from my main volume (three
                dedicated host servers with HDDs and SDD/NVM caching and
                log volumes o<span style="caret-color: rgb(0, 0, 0);">n
                  ZFS </span>). The cluster.shd-* entries are especially
                recommended. This is on gluster 9.4 at the moment, so
                some of these won’t map exactly.
                <div><br>
                </div>
                <div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">Volume Name: gv1</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">Type: Replicate</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);">Number
                    of Bricks: 1 x 3 = 3</div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">Transport-type: tcp</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);">Options
                    Reconfigured:</div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">cluster.read-hash-mode:
                      3</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.client-io-threads:
                      on</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.write-behind-window-size:
                      64MB</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.cache-size:
                      1G</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);">nfs.disable:
                    on</div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.readdir-ahead:
                      on</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.quick-read:
                      off</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.read-ahead:
                      on</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.io-cache:
                      off</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.stat-prefetch:
                      on</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">cluster.eager-lock:
                      enable</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">network.remote-dio:
                      enable</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);">server.event-threads:
                    4</div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">client.event-threads:
                      8</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.io-thread-count:
                      64</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">performance.low-prio-threads:
                      32</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">features.shard: on</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">features.shard-block-size:
                      64MB</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">cluster.locking-scheme:
                      granular</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);">cluster.data-self-heal-algorithm:
                    full</div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">cluster.shd-max-threads:
                      8</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">cluster.shd-wait-qlength:
                      10240</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">cluster.choose-local:
                      false</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><span
style="font-variant-ligatures: no-common-ligatures">cluster.granular-entry-heal:
                      enable</span></div>
                  <div
style="margin: 0px; font-stretch: normal; font-size: 15px; line-height: normal; font-family: "Fantasque Sans Mono"; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal; color: rgb(40, 255, 0); background-color: rgba(255, 255, 255, 0.9);"><br>
                  </div>
                  <div>Otherwise, more details about your servers, CPU,
                    RAM, and Disks would be useful for suggestions, and
                    details of your network as well. And if you haven’t
                    done kernel level tuning on the servers, you should
                    address that as well. These all vary a lot by your
                    work load and hardware setup, so there aren’t many
                    generic recommendations I can give other than to
                    make sure you tuned your tcp stack and enabled the
                    none disk elevator on SSDs or disks used by ZFS. </div>
                  <div><br>
                  </div>
                  <div>There’s a lot of tuning suggesting in the
                    archives if you go searching as well.</div>
                  <div><br>
                  </div>
                  <div>  -Darrell</div>
                  <div><br>
                  </div>
                  <div><br>
                    <blockquote type="cite">
                      <div>On Apr 9, 2024, at 3:05 AM, Ilias Chasapakis
                        forumZFD <a class="moz-txt-link-rfc2396E"
                          href="mailto:chasapakis@forumZFD.de"
                          moz-do-not-send="true"><chasapakis@forumZFD.de></a>
                        wrote:</div>
                      <br class="Apple-interchange-newline">
                      <div>
                        <div>Dear all,<br>
                          <br>
                          we would like to describe the situation that
                          we have and that does not solve since a long
                          time, that means after many minor<br>
                          and major upgrades of GlusterFS<br>
                          <br>
                          We use a KVM environment for VMs for glusterfs
                          and host servers are updated regularly. Hosts
                          are disomogeneous hardware,<br>
                          but configured with same characteristics.<br>
                          <br>
                          The VMs have been also harmonized to use the
                          virtio drivers where available for devices and
                          resources reserved are the same<br>
                          on each host.<br>
                          <br>
                          Physical switch for hosts has been substituted
                          with a reliable one.<br>
                          <br>
                          Probing peers has been and is quite quick in
                          the heartbeat network and communication
                          between the servers for apparently has no
                          issues on disruptions.<br>
                          <br>
                          And I say apparently because what we have is:<br>
                          <br>
                          - always pending failed heals that used to
                          resolve by a rotated reboot of the gluster vms
                          (replica 3). Restarting only<br>
                          glusterfs related services (daemon, events
                          etc.) has no effect, only reboot brings
                          results<br>
                          - very often failed heals are directories<br>
                          <br>
                          We lately removed a brick that was on a vm on
                          a host that has been entirely substituted.
                          Re-added the brick, sync went on and<br>
                          all data was eventually synced and started
                          with 0 pending failed heals. Now it develops
                          failed heals too like its fellow<br>
                          bricks. Please take into account we healed all
                          the failed entries (manually with various
                          methods) before adding the third brick.<br>
                          <br>
                          After some days of operating, the count of
                          failed heals rises again, not really fast but
                          with new entries for sure (which might solve<br>
                          with rotated reboots, or not).<br>
                          <br>
                          We have gluster clients also on ctdbs that
                          connect to the gluster and mount via glusterfs
                          client. Windows roaming profiles shared via
                          smb become frequently corrupted,(they are
                          composed of a great number small files and are
                          though of big total dimension). Gluster nodes
                          are formatted with xfs.<br>
                          <br>
                          Also what we observer is that mounting with
                          the vfs option in smb on the ctdbs has some
                          kind of delay. This means that you can see the
                          shared folder on for example<br>
                          a Windows client machine on a ctdb, but not on
                          another ctdb in the cluster and then after a
                          while it appears there too. And this
                          frequently st<br>
                          <br>
                          <br>
                          This is an excerpt of entries on our shd logs:<br>
                          <br>
                          <blockquote type="cite">2024-04-08
                            10:13:26.213596 +0000] I [MSGID: 108026]
                            [afr-self-heal-entry.c:1080:afr_selfheal_entry_do]
                            0-gv-ho-replicate-0: performing full entry
                            selfheal on
                            2c621415-6223-4b66-a4ca-3f6f267a448d<br>
                            [2024-04-08 10:14:08.135911 +0000] W [MSGID:
                            114031]
                            [client-rpc-fops_v2.c:2457:client4_0_link_cbk]
                            0-gv-ho-client-5: remote operation failed.
                            [{source=<gfid:91d83f0e-1864-4ff3-9174-b7c956e20596>},
                            {target=(null)}, {errno=116},
                            {error=Veraltete Dateizugriffsnummer (file
                            handle)}]<br>
                            [2024-04-08 10:15:59.135908 +0000] W [MSGID:
                            114061]
                            [client-common.c:2992:client_pre_readdir_v2]
                            0-gv-ho-client-5: remote_fd is -1. EBADFD
                            [{gfid=6b5e599e-c836-4ebe-b16a-8224425b88c7},
                            {errno=77}, {error=Die Dateizugriffsnummer
                            ist in schlechter Verfassung}]<br>
                            [2024-04-08 10:30:25.013592 +0000] I [MSGID:
                            108026]
                            [afr-self-heal-entry.c:1080:afr_selfheal_entry_do]
                            0-gv-ho-replicate-0: performing full entry
                            selfheal on
                            24e82e12-5512-4679-9eb3-8bd098367db7<br>
                            [2024-04-08 10:33:17.613594 +0000] W [MSGID:
                            114031]
                            [client-rpc-fops_v2.c:2457:client4_0_link_cbk]
                            0-gv-ho-client-5: remote operation failed.
                            [{source=<gfid:ef9068fc-a329-4a21-88d2-265ecd3d208c>},
                            {target=(null)}, {errno=116},
                            {error=Veraltete Dateizugriffsnummer (file
                            handle)}]<br>
                            [2024-04-08 10:33:21.201359 +0000] W [MSGID:
                            114031]
                            [client-rpc-fops_v2.c:2457:client4_0_link_cbk]
                            0-gv-ho-client-5: remote operation failed.
                            [{source=<br>
                          </blockquote>
                          <br>
                          How are he clients mapped to real hosts in
                          order to know on which one´s logs to look at?<br>
                          <br>
                          We would like to go by exclusion to finally
                          eradicate this, possibly in a conservative way
                          (not rebuilding everything) and we<br>
                          <br>
                          are becoming clueless as to where to look at
                          as we also tried various options settings
                          regarding performance etc.<br>
                          <br>
                          Here is the set on our main volume:<br>
                          <br>
                          <blockquote type="cite">cluster.lookup-unhashed                 
                            on (DEFAULT)<br>
                            cluster.lookup-optimize                  on
                            (DEFAULT)<br>
                            cluster.min-free-disk                    10%
                            (DEFAULT)<br>
                            cluster.min-free-inodes                  5%
                            (DEFAULT)<br>
                            cluster.rebalance-stats                  off
                            (DEFAULT)<br>
                            cluster.subvols-per-directory           
                            (null) (DEFAULT)<br>
                            cluster.readdir-optimize                 off
                            (DEFAULT)<br>
                            cluster.rsync-hash-regex                
                            (null) (DEFAULT)<br>
                            cluster.extra-hash-regex                
                            (null) (DEFAULT)<br>
                            cluster.dht-xattr-name                  
                            trusted.glusterfs.dht (DEFAULT)<br>
                            cluster.randomize-hash-range-by-gfid     off
                            (DEFAULT)<br>
                            cluster.rebal-throttle                  
                            normal (DEFAULT)<br>
                            cluster.lock-migration off<br>
                            cluster.force-migration off<br>
                            cluster.local-volume-name               
                            (null) (DEFAULT)<br>
                            cluster.weighted-rebalance               on
                            (DEFAULT)<br>
                            cluster.switch-pattern                  
                            (null) (DEFAULT)<br>
                            cluster.entry-change-log                 on
                            (DEFAULT)<br>
                            cluster.read-subvolume                  
                            (null) (DEFAULT)<br>
                            cluster.read-subvolume-index             -1
                            (DEFAULT)<br>
                            cluster.read-hash-mode                   1
                            (DEFAULT)<br>
                            cluster.background-self-heal-count       8
                            (DEFAULT)<br>
                            cluster.metadata-self-heal on<br>
                            cluster.data-self-heal on<br>
                            cluster.entry-self-heal on<br>
                            cluster.self-heal-daemon enable<br>
                            cluster.heal-timeout                     600
                            (DEFAULT)<br>
                            cluster.self-heal-window-size            8
                            (DEFAULT)<br>
                            cluster.data-change-log                  on
                            (DEFAULT)<br>
                            cluster.metadata-change-log              on
                            (DEFAULT)<br>
                            cluster.data-self-heal-algorithm        
                            (null) (DEFAULT)<br>
                            cluster.eager-lock                       on
                            (DEFAULT)<br>
                            disperse.eager-lock                      on
                            (DEFAULT)<br>
                            disperse.other-eager-lock                on
                            (DEFAULT)<br>
                            disperse.eager-lock-timeout              1
                            (DEFAULT)<br>
                            disperse.other-eager-lock-timeout        1
                            (DEFAULT)<br>
                            cluster.quorum-type auto<br>
                            cluster.quorum-count 2<br>
                            cluster.choose-local                    
                            true (DEFAULT)<br>
                            cluster.self-heal-readdir-size           1KB
                            (DEFAULT)<br>
                            cluster.post-op-delay-secs               1
                            (DEFAULT)<br>
                            cluster.ensure-durability                on
                            (DEFAULT)<br>
                            cluster.consistent-metadata              no
                            (DEFAULT)<br>
                            cluster.heal-wait-queue-length           128
                            (DEFAULT)<br>
                            cluster.favorite-child-policy none<br>
                            cluster.full-lock                        yes
                            (DEFAULT)<br>
                            cluster.optimistic-change-log            on
                            (DEFAULT)<br>
                            diagnostics.latency-measurement off<br>
                            diagnostics.dump-fd-stats                off
                            (DEFAULT)<br>
                            diagnostics.count-fop-hits off<br>
                            diagnostics.brick-log-level INFO<br>
                            diagnostics.client-log-level INFO<br>
                            diagnostics.brick-sys-log-level         
                            CRITICAL (DEFAULT)<br>
                            diagnostics.client-sys-log-level        
                            CRITICAL (DEFAULT)<br>
                            diagnostics.brick-logger                
                            (null) (DEFAULT)<br>
                            diagnostics.client-logger               
                            (null) (DEFAULT)<br>
                            diagnostics.brick-log-format            
                            (null) (DEFAULT)<br>
                            diagnostics.client-log-format           
                            (null) (DEFAULT)<br>
                            diagnostics.brick-log-buf-size           5
                            (DEFAULT)<br>
                            diagnostics.client-log-buf-size          5
                            (DEFAULT)<br>
                            diagnostics.brick-log-flush-timeout      120
                            (DEFAULT)<br>
                            diagnostics.client-log-flush-timeout     120
                            (DEFAULT)<br>
                            diagnostics.stats-dump-interval          0
                            (DEFAULT)<br>
                            diagnostics.fop-sample-interval          0
                            (DEFAULT)<br>
                            diagnostics.stats-dump-format           
                            json (DEFAULT)<br>
                            diagnostics.fop-sample-buf-size         
                            65535 (DEFAULT)<br>
                            diagnostics.stats-dnscache-ttl-sec      
                            86400 (DEFAULT)<br>
                            performance.cache-max-file-size 10<br>
                            performance.cache-min-file-size          0
                            (DEFAULT)<br>
                            performance.cache-refresh-timeout        1
                            (DEFAULT)<br>
                            performance.cache-priority (DEFAULT)<br>
                            performance.io-cache-size               
                            32MB (DEFAULT)<br>
                            performance.cache-size                  
                            32MB (DEFAULT)<br>
                            performance.io-thread-count              16
                            (DEFAULT)<br>
                            performance.high-prio-threads            16
                            (DEFAULT)<br>
                            performance.normal-prio-threads          16
                            (DEFAULT)<br>
                            performance.low-prio-threads             16
                            (DEFAULT)<br>
                            performance.least-prio-threads           1
                            (DEFAULT)<br>
                            performance.enable-least-priority        on
                            (DEFAULT)<br>
                            performance.iot-watchdog-secs           
                            (null) (DEFAULT)<br>
                            performance.iot-cleanup-disconnected-reqs
                            off (DEFAULT)<br>
                            performance.iot-pass-through            
                            false (DEFAULT)<br>
                            performance.io-cache-pass-through       
                            false (DEFAULT)<br>
                            performance.quick-read-cache-size       
                            128MB (DEFAULT)<br>
                            performance.cache-size                  
                            128MB (DEFAULT)<br>
                            performance.quick-read-cache-timeout     1
                            (DEFAULT)<br>
                            performance.qr-cache-timeout 600<br>
                            performance.quick-read-cache-invalidation
                            false (DEFAULT)<br>
                            performance.ctime-invalidation          
                            false (DEFAULT)<br>
                            performance.flush-behind                 on
                            (DEFAULT)<br>
                            performance.nfs.flush-behind             on
                            (DEFAULT)<br>
                            performance.write-behind-window-size 4MB<br>
                            performance.resync-failed-syncs-after-fsync
                            off (DEFAULT)<br>
                            performance.nfs.write-behind-window-size 1MB
                            (DEFAULT)<br>
                            performance.strict-o-direct              off
                            (DEFAULT)<br>
                            performance.nfs.strict-o-direct          off
                            (DEFAULT)<br>
                            performance.strict-write-ordering        off
                            (DEFAULT)<br>
                            performance.nfs.strict-write-ordering    off
                            (DEFAULT)<br>
                            performance.write-behind-trickling-writes on
                            (DEFAULT)<br>
                            performance.aggregate-size              
                            128KB (DEFAULT)<br>
performance.nfs.write-behind-trickling-writes on (DEFAULT)<br>
                            performance.lazy-open                    yes
                            (DEFAULT)<br>
                            performance.read-after-open              yes
                            (DEFAULT)<br>
                            performance.open-behind-pass-through    
                            false (DEFAULT)<br>
                            performance.read-ahead-page-count        4
                            (DEFAULT)<br>
                            performance.read-ahead-pass-through     
                            false (DEFAULT)<br>
                            performance.readdir-ahead-pass-through  
                            false (DEFAULT)<br>
                            performance.md-cache-pass-through       
                            false (DEFAULT)<br>
                            performance.write-behind-pass-through   
                            false (DEFAULT)<br>
                            performance.md-cache-timeout 600<br>
                            performance.cache-swift-metadata        
                            false (DEFAULT)<br>
                            performance.cache-samba-metadata on<br>
                            performance.cache-capability-xattrs     
                            true (DEFAULT)<br>
                            performance.cache-ima-xattrs            
                            true (DEFAULT)<br>
                            performance.md-cache-statfs              off
                            (DEFAULT)<br>
                            performance.xattr-cache-list (DEFAULT)<br>
                            performance.nl-cache-pass-through       
                            false (DEFAULT)<br>
                            network.frame-timeout                   
                            1800 (DEFAULT)<br>
                            network.ping-timeout 20<br>
                            network.tcp-window-size                 
                            (null) (DEFAULT)<br>
                            client.ssl off<br>
                            network.remote-dio                      
                            disable (DEFAULT)<br>
                            client.event-threads 4<br>
                            client.tcp-user-timeout 0<br>
                            client.keepalive-time 20<br>
                            client.keepalive-interval 2<br>
                            client.keepalive-count 9<br>
                            client.strict-locks off<br>
                            network.tcp-window-size                 
                            (null) (DEFAULT)<br>
                            network.inode-lru-limit 200000<br>
                            auth.allow *<br>
                            auth.reject                             
                            (null) (DEFAULT)<br>
                            transport.keepalive 1<br>
                            server.allow-insecure                    on
                            (DEFAULT)<br>
                            server.root-squash                       off
                            (DEFAULT)<br>
                            server.all-squash                        off
                            (DEFAULT)<br>
                            server.anonuid                          
                            65534 (DEFAULT)<br>
                            server.anongid                          
                            65534 (DEFAULT)<br>
                            server.statedump-path                   
                            /var/run/gluster (DEFAULT)<br>
                            server.outstanding-rpc-limit             64
                            (DEFAULT)<br>
                            server.ssl off<br>
                            auth.ssl-allow *<br>
                            server.manage-gids                       off
                            (DEFAULT)<br>
                            server.dynamic-auth                      on
                            (DEFAULT)<br>
                            client.send-gids                         on
                            (DEFAULT)<br>
                            server.gid-timeout                       300
                            (DEFAULT)<br>
                            server.own-thread                       
                            (null) (DEFAULT)<br>
                            server.event-threads 4<br>
                            server.tcp-user-timeout                  42
                            (DEFAULT)<br>
                            server.keepalive-time 20<br>
                            server.keepalive-interval 2<br>
                            server.keepalive-count 9<br>
                            transport.listen-backlog 1024<br>
                            ssl.own-cert                            
                            (null) (DEFAULT)<br>
                            ssl.private-key                         
                            (null) (DEFAULT)<br>
                            ssl.ca-list                             
                            (null) (DEFAULT)<br>
                            ssl.crl-path                            
                            (null) (DEFAULT)<br>
                            ssl.certificate-depth                   
                            (null) (DEFAULT)<br>
                            ssl.cipher-list                         
                            (null) (DEFAULT)<br>
                            ssl.dh-param                            
                            (null) (DEFAULT)<br>
                            ssl.ec-curve                            
                            (null) (DEFAULT)<br>
                            transport.address-family inet<br>
                            performance.write-behind off<br>
                            performance.read-ahead on<br>
                            performance.readdir-ahead on<br>
                            performance.io-cache off<br>
                            performance.open-behind on<br>
                            performance.quick-read on<br>
                            performance.nl-cache on<br>
                            performance.stat-prefetch on<br>
                            performance.client-io-threads off<br>
                            performance.nfs.write-behind on<br>
                            performance.nfs.read-ahead off<br>
                            performance.nfs.io-cache off<br>
                            performance.nfs.quick-read off<br>
                            performance.nfs.stat-prefetch off<br>
                            performance.nfs.io-threads off<br>
                            performance.force-readdirp              
                            true (DEFAULT)<br>
                            performance.cache-invalidation on<br>
                            performance.global-cache-invalidation   
                            true (DEFAULT)<br>
                            features.uss off<br>
                            features.snapshot-directory .snaps<br>
                            features.show-snapshot-directory off<br>
                            features.tag-namespaces off<br>
                            network.compression off<br>
                            network.compression.window-size          -15
                            (DEFAULT)<br>
                            network.compression.mem-level            8
                            (DEFAULT)<br>
                            network.compression.min-size             0
                            (DEFAULT)<br>
                            network.compression.compression-level    -1
                            (DEFAULT)<br>
                            network.compression.debug               
                            false (DEFAULT)<br>
                            features.default-soft-limit              80%
                            (DEFAULT)<br>
                            features.soft-timeout                    60
                            (DEFAULT)<br>
                            features.hard-timeout                    5
                            (DEFAULT)<br>
                            features.alert-time                     
                            86400 (DEFAULT)<br>
                            features.quota-deem-statfs off<br>
                            geo-replication.indexing off<br>
                            geo-replication.indexing off<br>
                            geo-replication.ignore-pid-check off<br>
                            geo-replication.ignore-pid-check off<br>
                            features.quota off<br>
                            features.inode-quota off<br>
                            features.bitrot disable<br>
                            debug.trace off<br>
                            debug.log-history                        no
                            (DEFAULT)<br>
                            debug.log-file                           no
                            (DEFAULT)<br>
                            debug.exclude-ops                       
                            (null) (DEFAULT)<br>
                            debug.include-ops                       
                            (null) (DEFAULT)<br>
                            debug.error-gen off<br>
                            debug.error-failure                     
                            (null) (DEFAULT)<br>
                            debug.error-number                      
                            (null) (DEFAULT)<br>
                            debug.random-failure                     off
                            (DEFAULT)<br>
                            debug.error-fops                        
                            (null) (DEFAULT)<br>
                            nfs.disable on<br>
                            features.read-only                       off
                            (DEFAULT)<br>
                            features.worm off<br>
                            features.worm-file-level off<br>
                            features.worm-files-deletable on<br>
                            features.default-retention-period        120
                            (DEFAULT)<br>
                            features.retention-mode                 
                            relax (DEFAULT)<br>
                            features.auto-commit-period              180
                            (DEFAULT)<br>
                            storage.linux-aio                        off
                            (DEFAULT)<br>
                            storage.linux-io_uring                   off
                            (DEFAULT)<br>
                            storage.batch-fsync-mode                
                            reverse-fsync (DEFAULT)<br>
                            storage.batch-fsync-delay-usec           0
                            (DEFAULT)<br>
                            storage.owner-uid                        -1
                            (DEFAULT)<br>
                            storage.owner-gid                        -1
                            (DEFAULT)<br>
                            storage.node-uuid-pathinfo               off
                            (DEFAULT)<br>
                            storage.health-check-interval            30
                            (DEFAULT)<br>
                            storage.build-pgfid                      off
                            (DEFAULT)<br>
                            storage.gfid2path                        on
                            (DEFAULT)<br>
                            storage.gfid2path-separator              :
                            (DEFAULT)<br>
                            storage.reserve                          1
                            (DEFAULT)<br>
                            storage.health-check-timeout             20
                            (DEFAULT)<br>
                            storage.fips-mode-rchecksum on<br>
                            storage.force-create-mode               
                            0000 (DEFAULT)<br>
                            storage.force-directory-mode            
                            0000 (DEFAULT)<br>
                            storage.create-mask                     
                            0777 (DEFAULT)<br>
                            storage.create-directory-mask           
                            0777 (DEFAULT)<br>
                            storage.max-hardlinks                    100
                            (DEFAULT)<br>
                            features.ctime                           on
                            (DEFAULT)<br>
                            config.gfproxyd off<br>
                            cluster.server-quorum-type server<br>
                            cluster.server-quorum-ratio 51<br>
                            changelog.changelog                      off
                            (DEFAULT)<br>
                            changelog.changelog-dir                  {{
                            brick.path }}/.glusterfs/changelogs
                            (DEFAULT)<br>
                            changelog.encoding                      
                            ascii (DEFAULT)<br>
                            changelog.rollover-time                  15
                            (DEFAULT)<br>
                            changelog.fsync-interval                 5
                            (DEFAULT)<br>
                            changelog.changelog-barrier-timeout 120<br>
                            changelog.capture-del-path               off
                            (DEFAULT)<br>
                            features.barrier disable<br>
                            features.barrier-timeout 120<br>
                            features.trash                           off
                            (DEFAULT)<br>
                            features.trash-dir                      
                            .trashcan (DEFAULT)<br>
                            features.trash-eliminate-path           
                            (null) (DEFAULT)<br>
                            features.trash-max-filesize              5MB
                            (DEFAULT)<br>
                            features.trash-internal-op               off
                            (DEFAULT)<br>
                            cluster.enable-shared-storage disable<br>
                            locks.trace                              off
                            (DEFAULT)<br>
                            locks.mandatory-locking                  off
                            (DEFAULT)<br>
                            cluster.disperse-self-heal-daemon       
                            enable (DEFAULT)<br>
                            cluster.quorum-reads                     no
                            (DEFAULT)<br>
                            client.bind-insecure                    
                            (null) (DEFAULT)<br>
                            features.timeout                         45
                            (DEFAULT)<br>
                            features.failover-hosts                 
                            (null) (DEFAULT)<br>
                            features.shard off<br>
                            features.shard-block-size               
                            64MB (DEFAULT)<br>
                            features.shard-lru-limit                
                            16384 (DEFAULT)<br>
                            features.shard-deletion-rate             100
                            (DEFAULT)<br>
                            features.scrub-throttle lazy<br>
                            features.scrub-freq biweekly<br>
                            features.scrub                          
                            false (DEFAULT)<br>
                            features.expiry-time 120<br>
                            features.signer-threads 4<br>
                            features.cache-invalidation on<br>
                            features.cache-invalidation-timeout 600<br>
                            ganesha.enable off<br>
                            features.leases off<br>
                            features.lease-lock-recall-timeout       60
                            (DEFAULT)<br>
                            disperse.background-heals                8
                            (DEFAULT)<br>
                            disperse.heal-wait-qlength               128
                            (DEFAULT)<br>
                            cluster.heal-timeout                     600
                            (DEFAULT)<br>
                            dht.force-readdirp                       on
                            (DEFAULT)<br>
                            disperse.read-policy                    
                            gfid-hash (DEFAULT)<br>
                            cluster.shd-max-threads 4<br>
                            cluster.shd-wait-qlength                
                            1024 (DEFAULT)<br>
                            cluster.locking-scheme                  
                            full (DEFAULT)<br>
                            cluster.granular-entry-heal              no
                            (DEFAULT)<br>
                            features.locks-revocation-secs           0
                            (DEFAULT)<br>
                            features.locks-revocation-clear-all     
                            false (DEFAULT)<br>
                            features.locks-revocation-max-blocked    0
                            (DEFAULT)<br>
                            features.locks-monkey-unlocking         
                            false (DEFAULT)<br>
                            features.locks-notify-contention         yes
                            (DEFAULT)<br>
                            features.locks-notify-contention-delay   5
                            (DEFAULT)<br>
                            disperse.shd-max-threads                 1
                            (DEFAULT)<br>
                            disperse.shd-wait-qlength 4096<br>
                            disperse.cpu-extensions                 
                            auto (DEFAULT)<br>
                            disperse.self-heal-window-size           32
                            (DEFAULT)<br>
                            cluster.use-compound-fops off<br>
                            performance.parallel-readdir on<br>
                            performance.rda-request-size 131072<br>
                            performance.rda-low-wmark               
                            4096 (DEFAULT)<br>
                            performance.rda-high-wmark              
                            128KB (DEFAULT)<br>
                            performance.rda-cache-limit 10MB<br>
                            performance.nl-cache-positive-entry     
                            false (DEFAULT)<br>
                            performance.nl-cache-limit 10MB<br>
                            performance.nl-cache-timeout 600<br>
                            cluster.brick-multiplex disable<br>
                            cluster.brick-graceful-cleanup disable<br>
                            glusterd.vol_count_per_thread 100<br>
                            cluster.max-bricks-per-process 250<br>
                            disperse.optimistic-change-log           on
                            (DEFAULT)<br>
                            disperse.stripe-cache                    4
                            (DEFAULT)<br>
                            cluster.halo-enabled                    
                            False (DEFAULT)<br>
                            cluster.halo-shd-max-latency            
                            99999 (DEFAULT)<br>
                            cluster.halo-nfsd-max-latency            5
                            (DEFAULT)<br>
                            cluster.halo-max-latency                 5
                            (DEFAULT)<br>
                            cluster.halo-max-replicas               
                            99999 (DEFAULT)<br>
                            cluster.halo-min-replicas                2
                            (DEFAULT)<br>
                            features.selinux on<br>
                            cluster.daemon-log-level INFO<br>
                            debug.delay-gen off<br>
                            delay-gen.delay-percentage               10%
                            (DEFAULT)<br>
                            delay-gen.delay-duration                
                            100000 (DEFAULT)<br>
                            delay-gen.enable (DEFAULT)<br>
                            disperse.parallel-writes                 on
                            (DEFAULT)<br>
                            disperse.quorum-count                    0
                            (DEFAULT)<br>
                            features.sdfs off<br>
                            features.cloudsync off<br>
                            features.ctime on<br>
                            ctime.noatime on<br>
                            features.cloudsync-storetype            
                            (null) (DEFAULT)<br>
                            features.enforce-mandatory-lock off<br>
                            config.global-threading off<br>
                            config.client-threads 16<br>
                            config.brick-threads 16<br>
                            features.cloudsync-remote-read off<br>
                            features.cloudsync-store-id             
                            (null) (DEFAULT)<br>
                            features.cloudsync-product-id           
                            (null) (DEFAULT)<br>
                            features.acl enable<br>
                            cluster.use-anonymous-inode yes<br>
                            rebalance.ensure-durability              on
                            (DEFAULT)<br>
                          </blockquote>
                          <br>
                          Again, sorry for the long post. We would be
                          happy to have this solved as we are excited
                          using glusterfs and we would like to go back
                          to having a stable configuration.<br>
                          <br>
                          We always appreciate the spirit of
                          collaboration and reciprocal help on this
                          list.<br>
                          <br>
                          Best<br>
                          Ilias<br>
                          <br>
                          -- <br>
                          forumZFD<br>
                          Entschieden für Frieden | Committed to Peace<br>
                          <br>
                          Ilias Chasapakis<br>
                          Referent IT | IT Consultant<br>
                          <br>
                          Forum Ziviler Friedensdienst e.V. | Forum
                          Civil Peace Service<br>
                          Am Kölner Brett 8 | 50825 Köln | Germany<br>
                          <br>
                          Tel 0221 91273243 | Fax 0221 91273299 | <a
                            class="moz-txt-link-freetext"
                            href="http://www.forumzfd.de/"
                            moz-do-not-send="true">http://www.forumZFD.de</a><br>
                          <br>
                          Vorstand nach § 26 BGB,
                          einzelvertretungsberechtigt|Executive Board:<br>
                          Alexander Mauz, Sonja Wiekenberg-Mlalandle,
                          Jens von Bargen<br>
                          VR 17651 Amtsgericht Köln<br>
                          <br>
                          Spenden|Donations: IBAN DE90 4306 0967 4103
                          7264 00   BIC GENODEM1GLS<br>
                          <br>
                          ________<br>
                          <br>
                          <br>
                          <br>
                          Community Meeting Calendar:<br>
                          <br>
                          Schedule -<br>
                          Every 2nd and 4th Tuesday at 14:30 IST / 09:00
                          UTC<br>
                          Bridge: <a class="moz-txt-link-freetext"
                            href="https://meet.google.com/cpu-eiue-hvk"
                            moz-do-not-send="true">https://meet.google.com/cpu-eiue-hvk</a><br>
                          Gluster-users mailing list<br>
                          <a
class="moz-txt-link-abbreviated moz-txt-link-freetext"
                            href="mailto:Gluster-users@gluster.org"
                            moz-do-not-send="true">Gluster-users@gluster.org</a><br>
                          <a class="moz-txt-link-freetext"
href="https://lists.gluster.org/mailman/listinfo/gluster-users"
                            moz-do-not-send="true">https://lists.gluster.org/mailman/listinfo/gluster-users</a><br>
                        </div>
                      </div>
                    </blockquote>
                  </div>
                  <br>
                </div>
              </blockquote>
              <pre class="moz-signature" cols="72">-- 
forumZFD
Entschieden für Frieden | Committed to Peace

Ilias Chasapakis
Referent IT | IT Consultant

Forum Ziviler Friedensdienst e.V. | Forum Civil Peace Service
Am Kölner Brett 8 | 50825 Köln | Germany

Tel 0221 91273243 | Fax 0221 91273299 | <a class="moz-txt-link-freetext"
              href="http://www.forumzfd.de/" moz-do-not-send="true">http://www.forumZFD.de</a>

Vorstand nach § 26 BGB, einzelvertretungsberechtigt|Executive Board:
Alexander Mauz, Sonja Wiekenberg-Mlalandle, Jens von Bargen
VR 17651 Amtsgericht Köln

Spenden|Donations: IBAN DE90 4306 0967 4103 7264 00   BIC GENODEM1GLS</pre>
            </div>
            ________<br>
            <br>
            <br>
            <br>
            Community Meeting Calendar:<br>
            <br>
            Schedule -<br>
            Every 2nd and 4th Tuesday at 14:30 IST / 09:00 UTC<br>
            Bridge: <a class="moz-txt-link-freetext" href="https://meet.google.com/cpu-eiue-hvk">https://meet.google.com/cpu-eiue-hvk</a><br>
            Gluster-users mailing list<br>
            <a class="moz-txt-link-abbreviated" href="mailto:Gluster-users@gluster.org">Gluster-users@gluster.org</a><br>
            <a class="moz-txt-link-freetext" href="https://lists.gluster.org/mailman/listinfo/gluster-users">https://lists.gluster.org/mailman/listinfo/gluster-users</a><br>
          </div>
        </blockquote>
      </div>
      <br>
    </blockquote>
    <pre class="moz-signature" cols="72">-- 
forumZFD
Entschieden für Frieden | Committed to Peace

Ilias Chasapakis
Referent IT | IT Consultant

Forum Ziviler Friedensdienst e.V. | Forum Civil Peace Service
Am Kölner Brett 8 | 50825 Köln | Germany

Tel 0221 91273243 | Fax 0221 91273299 | <a class="moz-txt-link-freetext" href="http://www.forumZFD.de">http://www.forumZFD.de</a>

Vorstand nach § 26 BGB, einzelvertretungsberechtigt|Executive Board:
Alexander Mauz, Sonja Wiekenberg-Mlalandle, Jens von Bargen
VR 17651 Amtsgericht Köln

Spenden|Donations: IBAN DE90 4306 0967 4103 7264 00   BIC GENODEM1GLS</pre>
  </body>
</html>