Merge branch '21642-io-panel-collection-tab-bug' into main. Closes #21642
[arvados.git] / doc / install / configure-fs-storage.html.textile.liquid
index be0a48cb8ca6d5526d600bda92775860099f9548..ebf964a796ceaf06d18ced756999808af08af29e 100644 (file)
@@ -17,31 +17,17 @@ Volumes are configured in the @Volumes@ section of the cluster configuration fil
 
 {% include 'assign_volume_uuid' %}
 
-Note that each volume has an AccessViaHosts section indicating that (for example) keep0's /mnt/local-disk directory is volume 0, while keep1's /mnt/local-disk directory is volume 1.
+Note that each volume entry has an @AccessViaHosts@ section indicating which Keepstore instance(s) will serve that volume.  In this example, keep0 and keep1 each have their own data disk.  The @/mnt/local-disk@ directory on keep0 is volume @ClusterID-nyw5e-000000000000000@, and the @/mnt/local-disk@ directory on keep1 is volume @ClusterID-nyw5e-000000000000001@ .
 
 <notextile>
-<pre><code>Clusters:
-  <span class="userinput">uuid_prefix</span>:
-    Volumes:
-      <span class="userinput">uuid_prefix</span>-nyw5e-<span class="userinput">000000000000000</span>:
+<pre><code>    Volumes:
+      <span class="userinput">ClusterID</span>-nyw5e-<span class="userinput">000000000000000</span>:
         AccessViaHosts:
-          "http://<span class="userinput">keep0.uuid_prefix.example.com</span>:25107": {}
-        Driver: Directory
+          "http://<span class="userinput">keep0.ClusterID.example.com</span>:25107": {}
+        Driver: <span class="userinput">Directory</span>
         DriverParameters:
           # The directory that will be used as the backing store.
-          Root: /mnt/local-disk
-
-          # When true, read and write operations (for whole 64MiB
-          # blocks) on an individual volume will queued and issued
-          # serially.  When false, read and write operations will be
-          # issued concurrently.
-          #
-          # May improve throughput if you experience contention when
-          # there are multiple requests to the same volume.
-          #
-          # When using SSDs, RAID, or a shared network filesystem, you
-          # probably don't want this.
-          Serialize: false
+          Root: <span class="userinput">/mnt/local-disk</span>
 
         # How much replication is performed by the underlying
         # filesystem.  (for example, a network filesystem may provide
@@ -53,25 +39,22 @@ Note that each volume has an AccessViaHosts section indicating that (for example
         # reads.
         ReadOnly: false
 
-        # Storage classes to associate with this volume.  See "Storage
-        # classes" in the "Admin" section of doc.arvados.org.
+        # <a href="{{site.baseurl}}/admin/storage-classes.html">Storage classes</a> to associate with this volume.
         StorageClasses: null
 
-      <span class="userinput">uuid_prefix</span>-nyw5e-<span class="userinput">000000000000001</span>:
+      <span class="userinput">ClusterID</span>-nyw5e-<span class="userinput">000000000000001</span>:
         AccessViaHosts:
-          "http://keep1.<span class="userinput">uuid_prefix</span>.example.com:25107": {}
-        Driver: Directory
+          "http://<span class="userinput">keep1.ClusterID.example.com</span>:25107": {}
+        Driver: <span class="userinput">Directory</span>
         DriverParameters:
-          Root: /mnt/local-disk
+          Root: <span class="userinput">/mnt/local-disk</span>
 </code></pre></notextile>
 
-In the case of a network-attached filesystem, the AccessViaHosts section can have multiple entries. If the filesystem is accessible by all keepstore servers, the AccessViaHosts section can be empty, or omitted entirely.
+In the case of a network-attached filesystem, the @AccessViaHosts@ section can have multiple entries. If the filesystem is accessible by all keepstore servers, the AccessViaHosts section can be empty, or omitted entirely.  In this example, the underlying storage system performs replication, so specifying @Replication: 2@ means a block is considered to be stored twice for the purposes of data integrity, while only stored on a single volume from the perspective of Keep.
 
 <notextile>
-<pre><code>Clusters:
-  <span class="userinput">uuid_prefix</span>:
-    Volumes:
-      <span class="userinput">uuid_prefix</span>-nyw5e-<span class="userinput">000000000000002</span>:
+<pre><code>    Volumes:
+      <span class="userinput">ClusterID</span>-nyw5e-<span class="userinput">000000000000002</span>:
         AccessViaHosts:
           # This section determines which keepstore servers access the
           # volume. In this example, keep0 has read/write access, and
@@ -80,10 +63,10 @@ In the case of a network-attached filesystem, the AccessViaHosts section can hav
           # If the AccessViaHosts section is empty or omitted, all
           # keepstore servers will have read/write access to the
           # volume.
-          "http://<span class="userinput">keep0.uuid_prefix.example.com</span>:25107/": {}
-          "http://<span class="userinput">keep1.uuid_prefix.example.com</span>:25107/": {ReadOnly: true}
-        Driver: Directory
+          "http://<span class="userinput">keep0.ClusterID.example.com</span>:25107/": {}
+          "http://<span class="userinput">keep1.ClusterID.example.com</span>:25107/": {ReadOnly: true}
+        Driver: <span class="userinput">Directory</span>
         DriverParameters:
-          Root: /mnt/network-attached-filesystem
+          Root: <span class="userinput">/mnt/network-attached-filesystem</span>
         Replication: 2
 </code></pre></notextile>