Remove implication that transactions restart themselves after segment failure. (#7534)

08969b8f · Chuck Litzell · David Yozie · ee76756c · 08969b8f · 08969b8f
2 changed file
--- a/gpdb-doc/dita/admin_guide/highavail/topics/g-overview-of-segment-mirroring.xml
+++ b/gpdb-doc/dita/admin_guide/highavail/topics/g-overview-of-segment-mirroring.xml
@@ -8,8 +8,12 @@
            instances: <i>primary</i> and <i>mirror</i>. Each primary segment has one corresponding
            mirror segment. A primary segment instance receives requests from the master to make
            changes to the segment data and then replicates those changes to the corresponding
-            mirror. If a primary segment becomes unavailable, database queries fail over to the
-            mirror segment.</p>
+            mirror. If Greenplum Database detects that a primary segment has failed or become
+            unavailable, it changes the role of its mirror segment to primary segment and the role
+            of the unavailable primary segment to mirror segment. Transactions in progress when the
+            failure occurred roll back and must be restarted. The administrator must then recover
+            the mirror segment, allow the mirror to syncronize with the current primary segment, and
+            then exchange the primary and mirror segments so they are in their preferred roles.</p>
        <p>If segment mirroring is not enabled, the Greenplum Database system shuts down if a
            segment instance fails. Administrators must manually recover all failed segments before
            Greenplum Database operations can resume.</p>

--- a/gpdb-doc/dita/admin_guide/managing/highavail.xml
+++ b/gpdb-doc/dita/admin_guide/managing/highavail.xml
@@ -253,10 +253,13 @@ mirror3=3:sdw2:sdw2-2:52002:53002:54002:/gpdata/mir1/gp3
    <title id="ki155550">Detecting a Failed Segment</title>
    <body>
      <p>With mirroring enabled, Greenplum Database automatically fails over to a mirror segment
-        when a primary segment goes down. Provided one segment instance is online per portion of
-        data, users may not realize a segment is down. If a transaction is in progress when a fault
-        occurs, the in-progress transaction rolls back and restarts automatically on the
-        reconfigured set of segments.</p>
+        when a primary segment goes down. The mirror segment assumes the role of the primary segment
+        and the failed primary segment becomes the mirror. Transactions in progress when a fault
+        occurs roll back and must be restarted. When restarted the transactions run on the new
+        primary segment.</p>
+      <p>To return the Greenplum Database cluster to full redundancy and balance, an administrator
+        recovers the down mirror segment and exchanges the roles of the primary and mirror
+        segments so that both segments are in their preferred roles.</p>
      <p>If the entire Greenplum Database system becomes nonoperational due to a segment failure
        (for example, if mirroring is not enabled or not enough segments are online to access all
        user data), users will see errors when trying to connect to a database. The errors returned