1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
|
#!@BASH_SHELL@
#
# Copyright (C) 1997-2003 Sistina Software, Inc. All rights reserved.
# Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
#
# Do reclaim-broadcasts when we kill lockd during shutdown/startup
# of a cluster service.
#
# Exported functions:
#
# notify_list_store
# notify_list_merge
# notify_list_broadcast
#
#
# Usage:
# statd_notify <directory> <hostname|ip>
#
# Copy out a list from <directory>, merge them with the system nfs lock
# list, and send them out as <hostname|ip> after generating a random
# state (needed so clients will reclaim their locks)
#
nfslock_statd_notify()
{
declare tmpdir
declare nl_dir=$1
declare nl_ip=$2
declare command # Work around bugs in rpc.statd
declare pid_xxx # Work around bugs in rpc.statd
declare owner
[ -z "$lockd_pid" ] && return 0
if ! [ -d $nl_dir ]; then
return 0
fi
if [ -z "`ls $nl_dir/sm/* 2> /dev/null`" ]; then
ocf_log debug "No hosts to notify"
return 0
fi
tmpdir=$(mktemp -d /tmp/statd-$2.XXXXXX)
# Ok, copy the HA directory to something we can use.
mkdir -p $tmpdir/sm
# Copy in our specified entries
cp -f $nl_dir/sm/* $tmpdir/sm
# Copy in our global entries
# XXX This might be what we just copied.
if [ -d "/var/lib/nfs/statd/sm" ]; then
owner=$(ls -dl /var/lib/nfs/statd/sm | awk '{print $3"."$4}')
cp -f /var/lib/nfs/statd/sm/* $tmpdir/sm
elif [ -d "/var/lib/nfs/sm" ]; then
owner=$(ls -dl /var/lib/nfs/statd/sm | awk '{print $3"."$4}')
cp -f /var/lib/nfs/sm/* $tmpdir/sm
fi
#
# Generate a random state file. If this ends up being what a client
# already has in its list, that's bad, but the chances of this
# are small - and relocations should be rare.
#
dd if=/dev/urandom of=$tmpdir/state bs=1 count=4 &> /dev/null
#
# Make sure we set permissions, or statd will not like it.
#
chown -R $owner $tmpdir
#
# Tell rpc.statd to notify clients. Don't go into background,
# because statd is buggy and won't exit like it's supposed to after
# sending the notifications out.
#
ocf_log info "Sending reclaim notifications via $nl_ip"
command="rpc.statd -NFP $tmpdir -n $nl_ip"
eval $command 2>&1 &
sleep 3 # XXX - the instance of rpc.statd we just spawned is supposed
# to exit after it finishes notifying clients.
# rpc.statd spawned which is still running handles the actual
# new SM_MON requests... we hope 3 seconds is enough time
# to get all the SM_NOTIFY messages out. rpc.statd = bugged
#
# clean up
#
pid_xxx=`ps auwwx | grep "$command" | grep -v grep | awk '{print $2}'`
kill $pid_xxx
rm -rf $tmpdir
return 0
}
#
# Copy of isSlave from svclib_ip and/or ip.sh
#
nfslock_isSlave()
{
declare intf=$1
declare line
if [ -z "$intf" ]; then
ocf_log err "usage: isSlave <I/F>"
return 1
fi
line=$(/sbin/ip link list dev $intf)
if [ $? -ne 0 ]; then
ocf_log err "$intf not found"
return 1
fi
if [ "$line" = "${line/<*SLAVE*>/}" ]; then
return 2
fi
# Yes, it is a slave device. Ignore.
return 0
}
#
# Get all the IPs on the system except loopback IPs
#
nfslock_ip_address_list()
{
declare idx dev family ifaddr
while read idx dev family ifaddr; do
if [ "$family" != "inet" ] && [ "$family" != "inet6" ]; then
continue
fi
if [ "$dev" = "lo" ]; then
# Ignore loopback
continue
fi
nfslock_isSlave $dev
if [ $? -ne 2 ]; then
continue
fi
idx=${idx/:/}
echo $dev $family ${ifaddr/\/*/} ${ifaddr/*\//}
done < <(/sbin/ip -o addr list | awk '{print $1,$2,$3,$4}')
return 0
}
#
# Usage: broadcast_notify <state_directory>
#
# Send the contents of <state_directory> out via all IPs on the system.
#
notify_list_broadcast()
{
declare dev family addr maskbits ip_name
declare lockd_pid=$(pidof lockd)
declare nl_dir=$1
# First of all, send lockd a SIGKILL. We hope nfsd is running.
# If it is, this will cause lockd to reset the grace period for
# lock reclaiming.
if [ -n "$lockd_pid" ]; then
ocf_log info "Asking lockd to drop locks (pid $lockd_pid)"
kill -9 $lockd_pid
else
ocf_log warning "lockd not running; cannot notify clients"
return 1
fi
while read dev family addr maskbits; do
if [ "$family" != "inet" ]; then
continue
fi
ip_name=$(clufindhostname -i $addr)
if [ -z "$ip_name" ]; then
nfslock_statd_notify $nl_dir $addr
else
nfslock_statd_notify $nl_dir $ip_name
fi
done < <(nfslock_ip_address_list)
}
#
# Store the lock monitor list from rpc.statd - do this during a teardown
# after the IP addresses of a service have been taken offline. Note that
# this should be done by HA-callout programs, but this feature is not in
# RHEL3.
#
notify_list_store()
{
declare nl_dir=$1
declare owner
mkdir -p $nl_dir/sm
if [ -d "/var/lib/nfs/statd/sm" ]; then
if [ -z "`ls /var/lib/nfs/statd/sm/* 2> /dev/null`" ]; then
return 1
# nothing to do!
fi
owner=$(ls -dl /var/lib/nfs/statd/sm | awk '{print $3"."$4}')
cp -Rdpf /var/lib/nfs/statd/sm/* $nl_dir/sm
chown -R $owner $nl_dir
return 0
elif [ -d "/var/lib/nfs/sm" ]; then
if [ -z "`ls /var/lib/nfs/sm/* 2> /dev/null`" ]; then
return 1
# nothing to do!
fi
owner=$(ls -dl /var/lib/nfs/sm | awk '{print $3"."$4}')
cp -Rdpf /var/lib/nfs/sm/* $nl_dir/sm
chown -R $owner $nl_dir
return 0
fi
return 1
}
#
# Merge the contents of <nl_dir>/sm with the system-wide list
# Make sure ownership is right, or statd will hiccup. This should not
# actually ever be needed because statd will, upon getting a SM_MON
# request, create all the entries in this list. It's mostly for
# housekeeping for next time we relocate the service.
#
notify_list_merge()
{
declare nl_dir=$1
declare owner
if [ -z "`ls $nl_dir/* 2> /dev/null`" ]; then
return 1
fi
if [ -d "/var/lib/nfs/statd/sm" ]; then
owner=$(ls -dl /var/lib/nfs/statd/sm | awk '{print $3"."$4}')
cp -Rdpf $nl_dir/sm/* /var/lib/nfs/statd/sm
chown -R $owner $nl_dir
return 0
elif [ -d "/var/lib/nfs/sm" ]; then
owner=$(ls -dl /var/lib/nfs/sm | awk '{print $3"."$4}')
cp -Rdpf $nl_dir/sm/* /var/lib/nfs/sm
chown -R $owner $nl_dir
return 0
fi
return 1
}
|