Revision 7ea7bcf6 man/gnt-instance.sgml

b/man/gnt-instance.sgml
1828 1828
        <cmdsynopsis>
1829 1829
          <command>replace-disks</command>
1830 1830
          <arg>--submit</arg>
1831
          <arg>--early-release</arg>
1831 1832
          <arg choice="req">-p</arg>
1832 1833
          <arg>--disks <replaceable>idx</replaceable></arg>
1833 1834
          <arg choice="req"><replaceable>instance</replaceable></arg>
......
1836 1837
        <cmdsynopsis>
1837 1838
          <command>replace-disks</command>
1838 1839
          <arg>--submit</arg>
1840
          <arg>--early-release</arg>
1839 1841
          <arg choice="req">-s</arg>
1840 1842
          <arg>--disks <replaceable>idx</replaceable></arg>
1841 1843
          <arg choice="req"><replaceable>instance</replaceable></arg>
......
1844 1846
        <cmdsynopsis>
1845 1847
          <command>replace-disks</command>
1846 1848
          <arg>--submit</arg>
1849
          <arg>--early-release</arg>
1847 1850
          <group choice="req">
1848 1851
            <arg>--iallocator <replaceable>name</replaceable></arg>
1849 1852
            <arg>--new-secondary <replaceable>NODE</replaceable></arg>
......
1855 1858
        <cmdsynopsis>
1856 1859
          <command>replace-disks</command>
1857 1860
          <arg>--submit</arg>
1861
          <arg>--early-release</arg>
1858 1862
          <arg choice="req">--auto</arg>
1859 1863
          <arg choice="req"><replaceable>instance</replaceable></arg>
1860 1864
        </cmdsynopsis>
......
1906 1910
        </para>
1907 1911

  
1908 1912
        <para>
1913
          The <option>--early-release</option> changes the code so
1914
          that the old storage on secondary node(s) is removed early
1915
          (before the resync is completed) and the internal Ganeti
1916
          locks for the current (and new, if any) secondary node are
1917
          also released, thus allowing more parallelism in the cluster
1918
          operation. This should be used only when recovering from a
1919
          disk failure on the current secondary (thus the old storage
1920
          is already broken) or when the storage on the primary node
1921
          is known to be fine (thus we won't need the old storage for
1922
          potential recovery).
1923
        </para>
1924

  
1925
        <para>
1909 1926
          Note that it is not possible to select an offline or drained
1910 1927
          node as a new secondary.
1911 1928
        </para>

Also available in: Unified diff