1SBD(8)                       STONITH Block Device                       SBD(8)
2
3
4

NAME

6       sbd - STONITH Block Device daemon
7

SYNPOSIS

9       sbd [options] "command"
10

SUMMARY

12       The SBD (Storage-Based Death) daemon integrates with Pacemaker and a
13       watchdog device to arrange for nodes to reliably self-terminate when
14       fencing is required. SBD can be particularly useful in environments
15       where traditional fencing mechanisms are not possible.
16
17       This build of SBD has message exchanging via a shared block storage
18       disabled to solely serve the above purpose.
19
20   GENERAL OPTIONS
21       -v  Enable some verbose debug logging.
22
23       -h  Display a concise summary of "sbd" options.
24
25       -n node
26           Set local node name; defaults to "uname -n". This should not need
27           to be set.
28
29       -R  Do not enable realtime priority. By default, "sbd" runs at realtime
30           priority, locks itself into memory, and also acquires highest IO
31           priority to protect itself against interference from other
32           processes on the system. This is a debugging-only option.
33
34   watch
35       Example usage:
36
37               sbd -W -P watch
38
39       This command will make "sbd" start in daemon mode. It will constantly
40       monitor corosync/pacemaker_remoted and optionally take Pacemaker's
41       state into account.
42
43       "sbd" must be started on boot before the cluster stack! See below for
44       enabling this according to your boot environment.
45
46       The options for this mode are rarely specified directly on the
47       commandline directly, but most frequently set via /etc/sysconfig/sbd.
48
49       For this build withouth block device integration configuration of a
50       watchdog is mandatory.  The watchdog is activated at initial start of
51       the sbd daemon.
52
53       There is one "sbd" process that acts as a master to which all watchers
54       report; one for corosync/pacemaker_remote; and, optionally, one that
55       handles the Pacemaker integration.
56
57       -W  Enable or disable use of the system watchdog to protect against the
58           sbd processes failing and the node being left in an undefined
59           state. Specify this once to enable, twice to disable.
60
61           Defaults to enabled.
62
63       -w /dev/watchdog
64           This can be used to override the default watchdog device used and
65           should not usually be necessary.
66
67       -p /var/run/sbd.pid
68           This option can be used to specify a pidfile for the main sbd
69           process.
70
71       -F N
72           Number of failures before a failing servant process will not be
73           restarted immediately until the dampening delay has expired. If set
74           to zero, servants will be restarted immediately and indefinitely.
75           If set to one, a failed servant will be restarted once every -t
76           seconds. If set to a different value, the servant will be restarted
77           that many times within the dampening period and then delay.
78
79           Defaults to 1.
80
81       -t N
82           Dampening delay before faulty servants are restarted. Combined with
83           "-F 1", the most logical way to tune the restart frequency of
84           servant processes.  Default is 5 seconds.
85
86           If set to zero, processes will be restarted indefinitely and
87           immediately.
88
89       -P  Check Pacemaker quorum and node health.
90
91       -Z  Enable trace mode. Warning: this is unsafe for production, use at
92           your own risk! Specifying this once will turn all reboots or power-
93           offs, be they caused by self-fence decisions or messages, into a
94           crashdump.  Specifying this twice will just log them but not
95           continue running.
96
97       -T  By default, the daemon will set the watchdog timeout as specified
98           in the device metadata. However, this does not work for every
99           watchdog device.  In this case, you must manually ensure that the
100           watchdog timeout used by the system correctly matches the SBD
101           settings, and then specify this option to allow "sbd" to continue
102           with start-up.
103

Base system configuration

105   Configure a watchdog
106       The only purpose of this build of sbd is to give pacemaker access to a
107       hardware watchdog.  Thus it is mandatory that you configure your Linux
108       system to load a watchdog driver with hardware assistance (as is
109       available on most modern systems), such as hpwdt, iTCO_wdt, or others.
110       As a fall-back, you can use the softdog module.
111
112       No other software must access the watchdog timer; it can only be
113       accessed by one process at any given time. Some hardware vendors ship
114       systems management software that use the watchdog for system resets
115       (f.e. HP ASR daemon). Such software has to be disabled if the watchdog
116       is to be used by SBD.
117
118   Configure SBD to start on boot
119       On systems using "sysvinit", the "openais" or "corosync" system start-
120       up scripts must handle starting or stopping "sbd" as required before
121       starting the rest of the cluster stack.
122
123       For "systemd", sbd simply has to be enabled using
124
125               systemctl enable sbd.service
126
127       The daemon is brought online on each node before corosync and Pacemaker
128       are started, and terminated only after all other cluster components
129       have been shut down - ensuring that cluster resources are never
130       activated without SBD supervision.
131
132   Configuration via sysconfig
133       The system instance of "sbd" is configured via /etc/sysconfig/sbd.  In
134       this file, you must specify the watchdog-device used, as well as any
135       options to pass to the daemon:
136
137               SBD_WATCHDOG_DEV=/dev/watchdog
138               SBD_PACEMAKER="true"
139

Pacemaker CIB integration

141   General cluster properties
142       You must also enable STONITH in general, and set the stonith-watchdog-
143       timeout to be at least twice the timeout you have configured for the
144       hardware watchdog.  If that timeout is 5 seconds, this is a possible
145       configuration:
146
147               property stonith-watchdog-timeout="10s"
148

LICENSE

150       Copyright (C) 2008-2013 Lars Marowsky-Bree
151
152       Copyright (C) 2014-2016 Andrew Beekhof
153
154       This program is free software; you can redistribute it and/or modify it
155       under the terms of the GNU General Public License as published by the
156       Free Software Foundation; either version 2 of the License, or (at your
157       option) any later version.
158
159       This software is distributed in the hope that it will be useful, but
160       WITHOUT ANY WARRANTY; without even the implied warranty of
161       MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
162       General Public License for more details.
163
164       For details see the GNU General Public License at
165       http://www.gnu.org/licenses/gpl-2.0.html (version 2) and/or
166       http://www.gnu.org/licenses/gpl.html (the newest as per "any later").
167
168
169
170SBD                               2017-03-22                            SBD(8)
Impressum