1 <!doctype refentry PUBLIC "-//OASIS//DTD DocBook V4.1//EN" [
3 <!-- Please adjust the date whenever revising the manpage. -->
4 <!ENTITY dhdate "<date>June 08, 2010</date>">
5 <!-- SECTION should be 1-8, maybe w/ subsection other parameters are
6 allowed: see man(7), man(1). -->
7 <!ENTITY dhsection "<manvolnum>8</manvolnum>">
8 <!ENTITY dhucpackage "<refentrytitle>ganeti-masterd</refentrytitle>">
9 <!ENTITY dhpackage "ganeti-masterd">
11 <!ENTITY debian "<productname>Debian</productname>">
12 <!ENTITY gnu "<acronym>GNU</acronym>">
13 <!ENTITY gpl "&gnu; <acronym>GPL</acronym>">
14 <!ENTITY footer SYSTEM "footer.sgml">
23 <holder>Google Inc.</holder>
31 <refmiscinfo>Ganeti 2.2</refmiscinfo>
34 <refname>&dhpackage;</refname>
36 <refpurpose>Ganeti master daemon</refpurpose>
40 <command>&dhpackage; </command>
43 <arg>--no-voting</arg>
48 <title>DESCRIPTION</title>
51 The <command>&dhpackage;</command> is the daemon which is
52 responsible for the overall cluster coordination. Without it, no
53 change can be performed on the cluster.
57 For testing purposes, you can give the <option>-f</option>
58 option and the program won't detach from the running terminal.
62 Debug-level message can be activated by giving the
63 <option>-d</option> option.
68 The role of the master daemon is to coordinate all the actions
69 that change the state of the cluster. Things like accepting
70 new jobs, coordinating the changes on nodes (via RPC calls to
71 the respective node daemons), maintaining the configuration
72 and so on are done via this daemon.
76 The only action that can be done without the master daemon is
77 the failover of the master role to another node in the
78 cluster, via the <command>gnt-cluster
79 master-failover</command> command.
83 If the master daemon is stopped, the instances are not
84 affected, but they won't be restarted automatically in case of
90 <title>STARTUP</title>
92 At startup, the master daemon will confirm with the node
93 daemons that the node it is running is indeed the master node
94 of the cluster. It will abort if it doesn't get half plus one
95 positive answers (offline nodes are queried too, just in case
96 our configuration is stale).
100 For small clusters with a number of nodes down, and especially
101 for two-node clusters where the other has gone done, this
102 creates a problem. In this case the
103 <option>--no-voting</option> option can be used to skip this
104 process. The option requires interactive confirmation, as
105 having two masters on the same cluster is a very dangerous
106 situation and will most likely lead to data loss.
111 <title>JOB QUEUE</title>
113 The master daemon maintains a job queue (located under
115 class="directory">@LOCALSTATEDIR@/lib/ganeti/queue</filename>) in
116 which all current jobs are stored, one job per file serialized
117 in JSON format; in this directory a subdirectory called
118 <filename class="directory">archive</filename> holds archived
123 The moving of jobs from the current to the queue directory is
124 done via a request to the master; this can be accomplished
125 from the command line with the <command>gnt-job
126 archive</command> or <command>gnt-job autoarchive</command>
127 commands. In case of problems with the master, a job file can
128 simply be moved away or deleted (but this might leave the
129 cluster inconsistent).
135 <title>COMMUNICATION PROTOCOL</title>
137 The master accepts commands over a Unix socket, using JSON
138 serialized messages separated by a specific byte sequence. For
139 more details, see the design documentation supplied with
150 <!-- Keep this comment at the end of the file
155 sgml-minimize-attributes:nil
156 sgml-always-quote-attributes:t
159 sgml-parent-document:nil
160 sgml-default-dtd-file:nil
161 sgml-exposed-tags:nil
162 sgml-local-catalogs:nil
163 sgml-local-ecat-files:nil