3 # Run commands on CTDB nodes.
5 # See http://ctdb.samba.org/ for more information about CTDB.
7 # Copyright (C) Martin Schwenke 2008
9 # Based on an earlier script by Andrew Tridgell and Ronnie Sahlberg.
11 # Copyright (C) Andrew Tridgell 2007
13 # This program is free software; you can redistribute it and/or modify
14 # it under the terms of the GNU General Public License as published by
15 # the Free Software Foundation; either version 3 of the License, or
16 # (at your option) any later version.
18 # This program is distributed in the hope that it will be useful,
19 # but WITHOUT ANY WARRANTY; without even the implied warranty of
20 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
21 # GNU General Public License for more details.
23 # You should have received a copy of the GNU General Public License
24 # along with this program; if not, see <http://www.gnu.org/licenses/>.
31 Usage: onnode [OPTION] ... <NODES> <COMMAND> ...
33 -c Run in current working directory on specified nodes.
34 -o <prefix> Save standard output from each node to file <prefix>.<ip>
35 -p Run command in parallel on specified nodes.
36 -q Do not print node addresses (overrides -v).
37 -n Allow nodes to be specified by name.
38 -f Specify nodes file, overrides CTDB_NODES_FILE.
39 -v Print node address even for a single node.
40 <NODES> "all", "any", "ok" (or "healthy"), "con" (or "connected"),
41 "rm" (or "recmaster"), "lvs" (or "lvsmaster"),
42 "natgw" (or "natgwlist"); or
43 a node number (0 base); or
44 a hostname (if -n is specified); or
45 list (comma separated) of <NODES>; or
46 range (hyphen separated) of node numbers.
54 echo "Invalid <nodespec>" >&2 ; echo >&2
66 ctdb_base="${CTDB_BASE:-/etc/ctdb}"
70 # $POSIXLY_CORRECT means that the command passed to onnode can
71 # take options and getopt won't reorder things to make them
73 local temp=$(POSIXLY_CORRECT=1 getopt -n "$prog" -o "cf:hno:pqv" -l help -- "$@")
81 -c) current=true ; shift ;;
82 -f) CTDB_NODES_FILE="$2" ; shift 2 ;;
83 -n) names_ok=true ; shift ;;
84 -o) prefix="$2" ; shift 2 ;;
85 -p) parallel=true ; shift ;;
86 -q) quiet=true ; shift ;;
87 -v) verbose=true ; shift ;;
89 -h|--help|*) usage ;; # Shouldn't happen, so this is reasonable.
106 if [ -n "$node" -a "$node" != "#DEAD" ] ; then
109 echo "${prog}: \"node ${n}\" does not exist" >&2
116 # Subshell avoids hacks to restore $IFS.
121 *-*) seq "${i%-*}" "${i#*-}" 2>/dev/null || invalid_nodespec ;;
122 # Separate lines for readability.
123 all|any|ok|healthy|con|connected) echo "$i" ;;
124 rm|recmaster|lvs|lvsmaster|natgw|natgwlist) echo "$i" ;;
126 [ $i -gt -1 ] 2>/dev/null || $names_ok || invalid_nodespec
133 ctdb_status_output="" # cache
134 get_nodes_with_status ()
145 bits="0:[0-1]:[0-1]:[0-1]:[0-1]"
151 if [ -z "$ctdb_status_output" ] ; then
152 # FIXME: need to do something if $CTDB_NODES_SOCKETS is set.
153 ctdb_status_output=$(ctdb -Y status 2>/dev/null)
154 if [ $? -ne 0 ] ; then
155 echo "${prog}: unable to get status of CTDB nodes" >&2
158 ctdb_status_output="${ctdb_status_output#* }"
163 for i in $ctdb_status_output ; do
164 # Try removing bits from end.
165 local t="${i%:${bits}:}"
166 if [ "$t" != "$i" ] ; then
167 # Succeeded. Get address. NOTE: this is an optimisation.
168 # It might be better to get the node number and then get
169 # the nth node to get the address. This would make things
170 # more consistent if $ctdb_base/nodes actually contained
172 nodes="${nodes} ${t##*:}"
179 ctdb_props="" # cache
180 get_node_with_property ()
186 if [ "${ctdb_props##:${prop}:}" = "$ctdb_props" ] ; then
187 prop_node=$(ctdb "$prop" -Y 2>/dev/null)
188 # We only want the first line.
191 prop_node="${prop_node%%${nl}*}"
192 if [ $? -eq 0 ] ; then
193 ctdb_props="${ctdb_props}${ctdb_props:+ }:${prop}:${prop_node}"
198 prop_node="${ctdb_props##:${prop}:}"
199 prop_node="${prop_node%% *}"
201 if [ -n "$prop_node" ] ; then
202 echo_nth "$prop_node" $all_nodes
204 echo "${prog}: No ${prop} available" >&2
209 get_any_available_node ()
213 # We do a recursive onnode to find which nodes are up and running.
214 local out=$($0 -pq all ctdb pnn 2>&1)
217 local pnn="${line#PNN:}"
218 if [ "$pnn" != "$line" ] ; then
219 echo_nth "$pnn" $all_nodes
222 # Else must be an error message from a down node.
231 if [ -n "$CTDB_NODES_SOCKETS" ] ; then
232 all_nodes="$CTDB_NODES_SOCKETS"
234 [ -e "$CTDB_NODES_FILE" ] || CTDB_NODES_FILE="${ctdb_base}/nodes"
235 all_nodes=$(sed -e 's@#.*@@g' -e 's@ *@@g' -e 's@^$@#DEAD@' $CTDB_NODES_FILE)
240 for n in $(parse_nodespec "$1") ; do
241 [ $? != 0 ] && exit 1 # Required to catch exit in above subshell.
244 echo "${all_nodes//#DEAD/}"
247 get_any_available_node "$all_nodes" || exit 1
250 get_nodes_with_status "$all_nodes" "healthy" || exit 1
253 get_nodes_with_status "$all_nodes" "connected" || exit 1
256 get_node_with_property "$all_nodes" "recmaster" || exit 1
259 get_node_with_property "$all_nodes" "lvsmaster" || exit 1
262 get_node_with_property "$all_nodes" "natgwlist" || exit 1
264 [0-9]|[0-9][0-9]|[0-9][0-9][0-9])
265 echo_nth $n $all_nodes
268 $names_ok || invalid_nodespec
276 CTDB_SOCKET="$1" sh -c "$2" 3>/dev/null
281 if [ -n "$prefix" ] ; then
282 cat >"${prefix}.${n//\//_}"
283 elif $verbose && $parallel ; then
292 if $verbose && $parallel ; then
299 ######################################################################
303 $current && command="cd $PWD && $command"
306 if [ -n "$CTDB_NODES_SOCKETS" ] ; then
309 # Could "2>/dev/null || true" but want to see errors from typos in file.
310 [ -r "${ctdb_base}/onnode.conf" ] && . "${ctdb_base}/onnode.conf"
311 [ -n "$SSH" ] || SSH=ssh
312 if [ "$SSH" = "ssh" ] ; then
315 : # rsh? All bets are off!
319 ######################################################################
321 nodes=$(get_nodes "$nodespec")
322 [ $? != 0 ] && exit 1 # Required to catch exit in above subshell.
327 # If $nodes contains a space or a newline then assume multiple nodes.
330 [ "$nodes" != "${nodes%[ ${nl}]*}" ] && verbose=true
334 trap 'kill -TERM $pids 2>/dev/null' INT TERM
335 # There's a small race here where the kill can fail if no processes
336 # have been added to $pids and the script is interrupted. However,
337 # the part of the window where it matter is very small.
340 set -o pipefail 2>/dev/null
342 { exec 3>&1 ; { $SSH $ssh_opts $EXTRA_SSH_OPTS $n "$command" | stdout_filter >&3 ; } 2>&1 | stderr_filter ; } &
346 echo >&2 ; echo ">> NODE: $n <<" >&2
349 { exec 3>&1 ; { $SSH $ssh_opts $EXTRA_SSH_OPTS $n "$command" | stdout_filter >&3 ; } 2>&1 | stderr_filter ; }
350 [ $? = 0 ] || retcode=$?
357 [ $? = 0 ] || retcode=$?