1 <?xml version="1.0" encoding="iso-8859-1"?>
3 PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
4 "http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd">
6 <refentry id="ctdbd.1">
9 <refentrytitle>ctdbd</refentrytitle>
10 <manvolnum>1</manvolnum>
11 <refmiscinfo class="source">ctdb</refmiscinfo>
12 <refmiscinfo class="manual">CTDB - clustered TDB database</refmiscinfo>
16 <refname>ctdbd</refname>
17 <refpurpose>The CTDB cluster daemon</refpurpose>
22 <command>ctdbd</command>
23 <arg rep="repeat"><replaceable>OPTION</replaceable></arg>
28 <title>DESCRIPTION</title>
30 ctdbd is the main CTDB daemon.
34 Note that ctdbd is not usually invoked directly. It is invoked
35 via <citerefentry><refentrytitle>ctdbd_wrapper</refentrytitle>
36 <manvolnum>1</manvolnum></citerefentry> or via the initscript.
40 See <citerefentry><refentrytitle>ctdb</refentrytitle>
41 <manvolnum>7</manvolnum></citerefentry> for an overview of CTDB.
46 <title>GENERAL OPTIONS</title>
50 <term>-d, --debug=<parameter>DEBUGLEVEL</parameter></term>
53 This option sets the debug level to DEBUGLEVEL, which
54 controls what will be written by the logging
55 subsystem. The default is 0 which will only log important
56 events and errors. A larger number will provide additional
60 See the <citetitle>DEBUG LEVELS</citetitle> section in
61 <citerefentry><refentrytitle>ctdb</refentrytitle>
62 <manvolnum>7</manvolnum></citerefentry> for more
69 <term>--dbdir=<parameter>DIRECTORY</parameter></term>
72 DIRECTORY on local storage where ctdbd keeps a local copy of
73 TDB databases. This directory is local for each node and
74 should not be stored on the shared cluster filesystem.
77 This directory would usually be <filename>/var/lib/ctdb</filename>
83 <term>--dbdir-persistent=<parameter>DIRECTORY</parameter></term>
86 DIRECTORY on local storage where ctdbd keeps a local copy of
87 persistent TDB databases. This directory is local for each
88 node and should not be stored on the shared cluster
92 This directory would usually be
93 <filename>/var/lib/ctdb/persistent</filename>
99 <term>--dbdir-state=<parameter>DIRECTORY</parameter></term>
102 DIRECTORY on local storage where ctdbd keep internal state
103 TDB files. This directory is local for each node and
104 should not be stored on the shared cluster filesystem.
107 This directory would usually be
108 <filename>/var/lib/ctdb/state</filename>
114 <term>--event-script-dir=<parameter>DIRECTORY</parameter></term>
117 DIRECTORY where the CTDB event scripts are stored. See the
118 <citetitle>EVENT SCRIPTS</citetitle> section in
119 <citerefentry><refentrytitle>ctdb</refentrytitle>
120 <manvolnum>7</manvolnum></citerefentry> for more information.
123 Default is <envar>CTDB_BASE</envar>/events.d, so usually
124 <filename>/etc/ctdb/events.d</filename>, which is part of
125 the CTDB installation.
131 <term>--logging=<parameter>STRING</parameter></term>
134 STRING specifies where ctdbd will write its log. The
135 default is file:<filename>/var/log/log.ctdb</filename> or
136 similar - the prefix may differ depending on how CTDB was
144 <term>file:<parameter>FILENAME</parameter></term>
147 FILENAME where ctdbd will write its log. This is usually
148 <filename>/var/log/log.ctdb</filename>.
153 <term>syslog<optional>:<parameter>METHOD</parameter></optional></term>
156 CTDB will log to syslog. By default this will use
160 Under heavy loads syslog(3) can block if the syslog
161 daemon processes messages too slowly. This can
162 cause CTDB to block when logging.
165 If METHOD is specified then it specifies an
166 extension that causes logging to be done in a
167 non-blocking mode. Note that <emphasis>this may
168 cause messages to be dropped</emphasis>. METHOD
173 <term>nonblocking</term>
176 CTDB will log to syslog via
177 <filename>/dev/log</filename> in non-blocking
186 CTDB will log to syslog via UDP to
187 localhost:514. The syslog daemon must be
188 configured to listen on (at least)
189 localhost:514. Most syslog daemons will log
190 the messages with hostname "localhost" - this
191 is a limitation of the implementation, for
192 compatibility with more syslog daemons.
197 <term>udp-rfc5424</term>
200 As with "udp" but messages are sent in RFC5424
201 format. This method will log the correct
202 hostname but is not as widely implemented in
218 This option is used to activate the LVS capability on a CTDB
219 node. Please see the <citetitle>LVS</citetitle> section in
220 <citerefentry><refentrytitle>ctdb</refentrytitle>
221 <manvolnum>7</manvolnum></citerefentry> for more
228 <term>--max-persistent-check-errors=<parameter>NUM</parameter></term>
231 NUM specifies the maximum number of health check failures
232 allowed for persistent databases during startup.
235 The default value is 0. Setting this to non-zero allows a
236 node with unhealthy persistent databases to startup and
237 join the cluster as long as there is another node with
238 healthy persistent databases.
244 <term>--nlist=<parameter>FILENAME</parameter></term>
247 FILENAME containing a list of the private IP addresses, one
248 per line, for each node in the cluster. This file
249 <emphasis>must be the same on each node</emphasis> in the
253 Default is <envar>CTDB_BASE</envar>/nodes, so usually
254 <filename>/etc/ctdb/nodes</filename>.
260 <term>--no-lmaster</term>
263 This argument specifies that this node can NOT become an lmaster
264 for records in the database. This means that it will never show up
265 in the vnnmap. This feature is primarily used for making a cluster
266 span across a WAN link and use CTDB as a WAN-accelerator.
269 Please see the <citetitle>REMOTE CLUSTER NODES</citetitle>
270 section in <citerefentry><refentrytitle>ctdb</refentrytitle>
271 <manvolnum>7</manvolnum></citerefentry> for more
278 <term>--no-recmaster</term>
281 This argument specifies that this node can NOT become a recmaster
282 for the database. This feature is primarily used for making a cluster
283 span across a WAN link and use CTDB as a WAN-accelerator.
286 Please see the <citetitle>REMOTE CLUSTER NODES</citetitle>
287 section in <citerefentry><refentrytitle>ctdb</refentrytitle>
288 <manvolnum>7</manvolnum></citerefentry> for more
295 <term>--notification-script=<parameter>FILENAME</parameter></term>
298 FILENAME specifying a script to be invoked by ctdbd when
299 certain state changes occur.
303 <filename>/etc/ctdb/notify.sh</filename>.
306 Please see the <citetitle>NOTIFICATION SCRIPT</citetitle>
307 section in <citerefentry><refentrytitle>ctdb</refentrytitle>
308 <manvolnum>7</manvolnum></citerefentry> for more
315 <term>--pidfile=<parameter>FILENAME</parameter></term>
318 FILENAME for file containing process ID of main CTDB
319 daemon. This file is automatically created and removed by
323 The default is to not create a PID file.
329 <term>--public_addresses=<parameter>FILENAME</parameter></term>
332 FILENAME specifying a file containing the public IP
333 addresses to use on the cluster when CTDB should use IP
334 takeover. This file contains a list of IP addresses,
335 netmasks and interfaces. CTDB will distribute these public
336 IP addresses appropriately across the available nodes.
339 The IP addresses specified in this file can differ across
343 This is usually the file
344 <filename>/etc/ctdb/public_addresses</filename>
350 <term>--public-interface=<parameter>INTERFACE</parameter></term>
353 INTERFACE on which to attach public IP addresses or on which
354 to attach the single-public-ip when used.
357 When using public IP addresses, this is only required if
358 interfaces are not explicitly specified in the public
365 <term>--reclock=<parameter>FILENAME</parameter></term>
368 FILENAME is the name of the recovery lock file stored in
369 <emphasis>shared storage</emphasis> that ctdbd uses to
370 prevent split brains from occuring.
373 It is possible to run CTDB without a recovery lock file, but
374 then there will be no protection against split brain if the
375 cluster/network becomes partitioned. Using CTDB without a
376 reclock file is strongly discouraged.
382 <term>--single-public-ip=<parameter>IPADDR</parameter></term>
385 IPADDR specifies the single IP that CTDB will use in
389 Please see the <citetitle>LVS</citetitle> section in
390 <citerefentry><refentrytitle>ctdb</refentrytitle>
391 <manvolnum>7</manvolnum></citerefentry> for more
398 <term>--start-as-disabled</term>
401 This makes ctdbd start in the DISABLED state.
404 To allow the node to host public IP addresses and
405 services, it must be manually enabled using the
406 <command>ctdb enable</command> command.
409 Please see the <citetitle>NODE STATES</citetitle> section
410 in <citerefentry><refentrytitle>ctdb</refentrytitle>
411 <manvolnum>7</manvolnum></citerefentry> for more
412 information about the DISABLED state.
418 <term>--start-as-stopped</term>
421 This makes ctdbd start in the STOPPED state.
424 To allow the node to take part in the cluster it must be
425 manually continued with the the <command>ctdb
426 enable</command> command.
429 Please see the <citetitle>NODE STATES</citetitle> section
430 in <citerefentry><refentrytitle>ctdb</refentrytitle>
431 <manvolnum>7</manvolnum></citerefentry> for more
432 information about the STOPPED state.
438 <term>--syslog</term>
441 Send log messages to syslog instead of the CTDB logfile.
442 This option overrides --logfile. The default is to log to
449 <term>--transport=tcp|infiniband</term>
452 This option specifies which transport to use for ctdbd
453 internode communications. The default is "tcp".
456 The "infiniband" support is not regularly tested.
462 <term>-?, --help</term>
465 Display a summary of options.
474 <title>DEBUGGING OPTIONS</title>
479 <term>-i, --interactive</term>
482 Enable interactive mode. This will make ctdbd run in the
483 foreground and not detach from the terminal. By default
484 ctdbd will detach itself and run in the background as a
491 <term>--listen=<parameter>IPADDR</parameter></term>
494 This specifies which IP address that ctdbd will bind to.
497 By default ctdbd will bind to the first address it finds in
498 the <filename>/etc/ctdb/nodes</filename> file that is also
499 present on the local system.
502 This option is only required when you want to run multiple
503 ctdbd daemons/nodes on the same physical host in which case
504 there would be multiple entries in
505 <filename>/etc/ctdb/nodes</filename> that would match a
512 <term>--nopublicipcheck</term>
515 This option is used when testing with multiple local
516 daemons on a single machine. It disables checks related
517 to public IP addresses.
523 <term>--nosetsched</term>
526 This is a debugging option. This option is only used when
530 Normally ctdbd will change its scheduler to run as a
531 real-time process. This is the default mode for a normal
532 ctdbd operation to gurarantee that ctdbd always gets the CPU
533 cycles that it needs.
536 This option is used to tell ctdbd to
537 <emphasis>not</emphasis> run as a real-time process and
538 instead run ctdbd as a normal userspace process. This is
539 useful for debugging and when you want to run ctdbd under
540 valgrind or gdb. (You don't want to attach valgrind or gdb
541 to a real-time process.)
547 <term>--socket=<parameter>FILENAME</parameter></term>
550 FILENAME specifies the name of the Unix domain socket that
551 ctdbd will create. This socket is used by local clients to
552 communicate with ctdbd.
555 The default is <filename>/tmp/ctdb.socket</filename> . You
556 only need to use this option if you plan to run multiple
557 ctdbd daemons on the same physical host, usually for
564 <term>--script-log-level=<parameter>DEBUGLEVEL</parameter></term>
567 This option sets the debug level of event script output to
568 DEBUGLEVEL. The default is ERR (0).
571 See the <citetitle>DEBUG LEVELS</citetitle> section in
572 <citerefentry><refentrytitle>ctdb</refentrytitle>
573 <manvolnum>7</manvolnum></citerefentry> for more
580 <term>--sloppy-start</term>
583 This is debugging option. This speeds up the initial
584 recovery during startup at the expense of some consistency
585 checking. <emphasis>Don't use this option in
586 production</emphasis>.
592 <term>--torture</term>
595 This option is only used for development and testing of
596 CTDB. It adds artificial errors and failures to the
597 common codepaths in ctdbd to verify that ctdbd can recover
598 correctly from failures.
601 <emphasis>Do not use this option</emphasis> unless you are
602 developing and testing new functionality in CTDB.
608 <term>--valgrinding</term>
611 This is a debugging option. This option is only used when
612 debugging ctdbd. This enables additional debugging
613 capabilities and implies --nosetsched.
622 <title>SEE ALSO</title>
624 <citerefentry><refentrytitle>ctdb</refentrytitle>
625 <manvolnum>1</manvolnum></citerefentry>,
627 <citerefentry><refentrytitle>ctdbd_wrapper</refentrytitle>
628 <manvolnum>1</manvolnum></citerefentry>,
630 <citerefentry><refentrytitle>onnode</refentrytitle>
631 <manvolnum>1</manvolnum></citerefentry>,
633 <citerefentry><refentrytitle>ctdb</refentrytitle>
634 <manvolnum>7</manvolnum></citerefentry>,
636 <citerefentry><refentrytitle>ctdb-tunables</refentrytitle>
637 <manvolnum>7</manvolnum></citerefentry>,
639 <ulink url="http://ctdb.samba.org/"/>
646 This documentation was written by
655 <holder>Andrew Tridgell</holder>
656 <holder>Ronnie Sahlberg</holder>
660 This program is free software; you can redistribute it and/or
661 modify it under the terms of the GNU General Public License as
662 published by the Free Software Foundation; either version 3 of
663 the License, or (at your option) any later version.
666 This program is distributed in the hope that it will be
667 useful, but WITHOUT ANY WARRANTY; without even the implied
668 warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
669 PURPOSE. See the GNU General Public License for more details.
672 You should have received a copy of the GNU General Public
673 License along with this program; if not, see
674 <ulink url="http://www.gnu.org/licenses"/>.