sin/samba-mirror
1
0
Fork 0
mirror of https://github.com/samba-team/samba.git synced 2025-01-11 05:18:09 +03:00
Code

More untested eventscript factorisation.

Browse Source 
Signed-off-by: Martin Schwenke <martin@meltin.net>

(This used to be ctdb commit ac655b0a65b32d809d47fec9821f7f31bb2fe2a7)
This commit is contained in:
Martin Schwenke 2009-11-19 15:00:17 +11:00
parent 73cb65bf1a
commit ee513c1ba2
16 changed files with 160 additions and 240 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
ctdb/config/events.d/00.ctdb
View File

@ -10,13 +10,7 @@
# recovered : called when ctdb has finished a recovery event
. $CTDB_BASE/functions
loadconfig ctdb
# ensure we have /bin and /usr/bin in the path
PATH=/bin:/usr/bin:$PATH
cmd="$1"
shift
loadconfig
case $cmd in
startup)

58
ctdb/config/events.d/01.reclock
View File

@ -2,55 +2,43 @@
# script to check accessibility to the reclock file on a node
. $CTDB_BASE/functions
loadconfig ctdb
cmd="$1"
shift
PATH=/usr/bin:/bin:/usr/sbin:/sbin:$PATH
# Count the number of intervals that have passed when we have tried to
# but failed to stat the reclock file. after third failure the node
# becomes unhealthy after the twentieth failure the node we shutdown
# ctdbd
RECLOCKCOUNT="fail-count"
loadconfig
case $cmd in
startup)
ctdb_counter_init "$RECLOCKCOUNT"
startup)
ctdb_counter_init
;;
monitor)
ctdb_counter_incr "$RECLOCKCOUNT"
ctdb_counter_limit "$RECLOCKCOUNT" 200 && {
echo "Reclock file \"$RECLOCKFILE\" can not be accessed. Shutting down."
df
sleep 1
ctdb shutdown
monitor)
ctdb_counter_incr
(ctdb_counter_limit 200 >/dev/null 2>&1) || {
echo "Reclock file \"$RECLOCKFILE\" can not be accessed. Shutting down."
df
sleep 1
ctdb shutdown
}
RECLOCKFILE=`ctdb -Y getreclock`
RECLOCKFILE=$(ctdb -Y getreclock)
[ -z "$RECLOCKFILE" ] && {
# we are not using a reclock file
ctdb_counter_init "$RECLOCKCOUNT"
exit 0
# we are not using a reclock file
ctdb_counter_init
exit 0
}
# try stat the reclock file as a background process
# so that we dont block in case the cluster filesystem is unavailable
(
stat $RECLOCKFILE && {
# we could stat the file, reset the counter
ctdb_counter_init "$RECLOCKCOUNT"
}
stat $RECLOCKFILE && {
# we could stat the file, reset the counter
ctdb_counter_init
}
) >/dev/null 2>/dev/null &
ctdb_counter_limit "$RECLOCKCOUNT" 3 && {
echo "Reclock file \"$RECLOCKFILE\" can not be accessed. Mark node UNHEALTHY."
df
exit 1;
}
ctdb_counter_limit 3
;;
status)
ctdb_checkstatus || exit $?
;;
esac

11
ctdb/config/events.d/10.interface
View File

@ -6,10 +6,7 @@
# public interface
. $CTDB_BASE/functions
loadconfig ctdb
cmd="$1"
shift
loadconfig
[ -z "$CTDB_PUBLIC_ADDRESSES" ] && {
CTDB_PUBLIC_ADDRESSES=$CTDB_BASE/public_addresses
@ -177,10 +174,10 @@ case $cmd in
esac
done
;;
status)
ctdb_checkstatus || exit $?
;;
esac
exit 0

15
ctdb/config/events.d/11.natgw
View File

@ -6,15 +6,10 @@
#
. $CTDB_BASE/functions
loadconfig ctdb
loadconfig
[ -z "$CTDB_NATGW_PUBLIC_IFACE" ] && exit 0
cmd="$1"
shift
PATH=/usr/bin:/bin:/usr/sbin:/sbin:$PATH
delete_all() {
remove_ip $CTDB_NATGW_PUBLIC_IP $CTDB_NATGW_PUBLIC_IFACE
remove_ip $CTDB_NATGW_PUBLIC_IP_HOST lo
@ -28,7 +23,7 @@ delete_all() {
}
case $cmd in
startup)
startup)
# do not respond to ARPs that are for ip addresses with scope 'host'
echo 3 > /proc/sys/net/ipv4/conf/all/arp_ignore
# do not send out arp requests from loopback addresses
@ -37,13 +32,13 @@ case $cmd in
ctdb setnatgwstate on
;;
recovered|updatenatgw)
recovered|updatenatgw)
MYPNN=`ctdb pnn | cut -d: -f2`
NATGWMASTER=`ctdb natgwlist | head -1 | sed -e "s/ .*//"`
NATGWIP=`ctdb natgwlist | head -1 | sed -e "s/^[^ ]* *//"`
CTDB_NATGW_PUBLIC_IP_HOST=`echo $CTDB_NATGW_PUBLIC_IP | sed -e "s/\/.*/\/32/"`
if [ "$NATGWMASTER" = "-1" ]; then
if [ "$NATGWMASTER" == "-1" ]; then
echo "There is not NATGW master node"
exit 1
fi
@ -71,7 +66,7 @@ case $cmd in
echo 1 > /proc/sys/net/ipv4/route/flush
;;
shutdown|removenatgw)
shutdown|removenatgw)
delete_all
;;

6
ctdb/config/events.d/11.routing
View File

@ -13,16 +13,12 @@
# bond1 10.3.3.0/24 10.0.0.1
. $CTDB_BASE/functions
loadconfig ctdb
loadconfig
[ -f $CTDB_BASE/static-routes ] || {
exit 0
}
cmd="$1"
shift
PATH=/usr/bin:/bin:/usr/sbin:/sbin:$PATH
case $cmd in
takeip|releaseip)
iface=$1

8
ctdb/config/events.d/20.multipathd
View File

@ -6,14 +6,10 @@
# CTDB_MONITOR_MPDEVICES="device1 device2 ..."
#
PATH=/bin:/usr/bin:$PATH
service_name="multipathd"
. $CTDB_BASE/functions
loadconfig ctdb
loadconfig multipathd
cmd="$1"
shift
loadconfig
[ -z "$CTDB_MONITOR_MPDEVICES" ] && {
exit 0

41
ctdb/config/events.d/31.clamd
View File

@ -2,52 +2,45 @@
# event script to manage clamd in a cluster environment
. $CTDB_BASE/functions
loadconfig ctdb
detect_init_style
case $CTDB_INIT_STYLE in
redhat)
CTDB_SERVICE_CLAMD="clamd"
CTDB_CONFIG_CLAMD="clamd"
;;
suse)
CTDB_SERVICE_CLAMD="clamav"
CTDB_CONFIG_CLAMD="clamav"
;;
debian)
CTDB_SERVICE_CLAMD="clamav"
CTDB_CONFIG_CLAMD="clamav"
service_name="clamd"
service_config="clamd"
;;
*)
# should not happen.
# for now use red hat style as default
CTDB_SERVICE_CLAMD="clamd"
CTDB_CONFIG_CLAMD="clamd"
service_name="clamav"
service_config="clamav"
;;
esac
loadconfig "${CTDB_CONFIG_CLAMD}"
service_start="service $service_name stop > /dev/null 2>&1 ; service $service_name start"
service_stop="service $service_name stop"
[ "$CTDB_MANAGES_CLAMD" = "yes" ] || exit 0
loadconfig
cmd="$1"
shift
ctdb_start_stop_service
is_ctdb_managed_service || exit 0
case $cmd in
startup)
service "${CTDB_SERVICE_CLAMD}" stop > /dev/null 2>&1
service "${CTDB_SERVICE_CLAMD}" start
ctdb_service_start
;;
shutdown)
service "${CTDB_SERVICE_CLAMD}" stop
ctdb_service_stop
;;
monitor)
ctdb_check_unix_socket "clamd" ${CTDB_CLAMD_SOCKET}
ctdb_check_unix_socket ${CTDB_CLAMD_SOCKET} || exit $?
;;
status)
ctdb_checkstatus || exit $?
;;
esac
exit 0

5
ctdb/config/events.d/40.vsftpd
View File

@ -3,8 +3,8 @@
service_name="vsftpd"
# make sure the service is stopped first
service_start="service vsftpd stop > /dev/null 2>&1 ; service vsftpd start"
service_stop="service vsftpd stop"
service_start="service $service_name stop > /dev/null 2>&1 ; service $service_name start"
service_stop="service $service_name stop"
service_reconfigure="service $service_name restart"
service_fail_limit=2
service_tcp_ports=21
@ -54,6 +54,7 @@ case $cmd in
fi
fi
;;
status)
ctdb_checkstatus || exit $?
;;

6
ctdb/config/events.d/41.httpd
View File

@ -35,6 +35,8 @@ loadconfig
ctdb_start_stop_service
is_ctdb_managed_service || exit 0
case $cmd in
startup)
ctdb_service_start
@ -56,6 +58,10 @@ case $cmd in
exit 1
fi
;;
status)
ctdb_checkstatus || exit $?
;;
esac
exit 0

101
ctdb/config/events.d/60.nfs
View File

@ -1,71 +1,69 @@
#!/bin/sh
# script to manage nfs in a clustered environment
. $CTDB_BASE/functions
loadconfig ctdb
loadconfig nfs
[ "$CTDB_MANAGES_NFS" = "yes" ] || exit 0
[ -z "$STATD_SHARED_DIRECTORY" ] && exit 0
cmd="$1"
shift
PATH=/usr/bin:/bin:/usr/sbin:/sbin:$PATH
case $cmd in
startup)
start_nfs() {
/bin/mkdir -p $CTDB_BASE/state/nfs
/bin/mkdir -p $CTDB_BASE/state/statd/ip
/bin/mkdir -p $STATD_SHARED_DIRECTORY
# make sure nfs is stopped before we start it, or it may get a bind error
startstop_nfs stop
startstop_nfs start
;;
shutdown)
startstop_nfs stop
exit 0
;;
takeip)
ip=$2
echo $ip >> $CTDB_BASE/state/statd/restart
# having a list of what IPs we have allows statd to do the right
# thing via $CTDB_BASE/statd-callout
touch $CTDB_BASE/state/statd/ip/$ip
exit 0
;;
releaseip)
iface=$1
ip=$2
maskbits=$3
echo $ip >> $CTDB_BASE/state/statd/restart
/bin/rm -f $CTDB_BASE/state/statd/ip/$ip
exit 0
;;
recovered)
# if no IPs have changed then don't need to restart statd
[ -f $CTDB_BASE/state/statd/restart ] || exit 0;
}
reconfigure_nfs() {
# always restart the lockmanager so that we start with a clusterwide
# graceperiod when ip addresses has changed
[ -x $CTDB_BASE/statd-callout ] && {
$CTDB_BASE/statd-callout notify &
} >/dev/null 2>&1
/bin/rm -f $CTDB_BASE/state/statd/restart
}
. $CTDB_BASE/functions
service_name="nfs"
service_start="start_nfs"
service_stop="startstop_nfs stop"
service_reconfigure="reconfigure_nfs"
loadconfig
[ -z "$STATD_SHARED_DIRECTORY" ] && exit 0
ctdb_start_stop_service
case $cmd in
startup)
ctdb_service_start
;;
shutdown)
ctdb_service_stop
;;
takeip)
ctdb_service_set_reconfigure
touch $CTDB_BASE/state/statd/ip/$2
;;
releaseip)
ctdb_service_set_reconfigure
/bin/rm -f $CTDB_BASE/state/statd/ip/$2
;;
recovered)
# if we have taken or released any ips we must
# restart the lock manager so that we enter a clusterwide grace period
if ctdb_service_needs_reconfigure ; then
ctdb_service_reconfigure
fi
;;
monitor)
if ctdb_service_needs_reconfigure ; then
ctdb_service_reconfigure
exit 0
fi
# check that statd responds to rpc requests
# if statd is not running we try to restart it
rpcinfo -u localhost 100024 1 > /dev/null || {
@ -103,6 +101,9 @@ case $cmd in
}
;;
status)
ctdb_checkstatus || exit $?
;;
esac
exit 0

47
ctdb/config/events.d/61.nfstickle
View File

@ -1,25 +1,21 @@
#!/bin/sh
# ctdb event script for NFS tickle acks
PATH=/bin:/usr/bin:$PATH
. $CTDB_BASE/functions
loadconfig ctdb
loadconfig nfs
cmd="$1"
shift
service_name="nfs"
service_start="mkdir -p $CTDB_BASE/state/nfstickle;mkdir -p $NFS_TICKLE_SHARED_DIRECTORY/`hostname`;echo 1 > /proc/sys/net/ipv4/tcp_tw_recycle"
service_reconfigure=$service_start
loadconfig
ctdb_start_stop_service
[ "$CTDB_MANAGES_NFS" = "yes" ] || exit 0
[ -z "$NFS_TICKLE_SHARED_DIRECTORY" ] && exit 0
case $cmd in
startup)
mkdir -p $CTDB_BASE/state/nfstickle
mkdir -p $NFS_TICKLE_SHARED_DIRECTORY/`hostname`
# we rely on fast tcp wait1 recycling
echo 1 > /proc/sys/net/ipv4/tcp_tw_recycle
exit 0
ctdb_service_start
;;
takeip)
@ -31,46 +27,25 @@ case $cmd in
# send tickle acks for all the connections the old server had
for f in $NFS_TICKLE_SHARED_DIRECTORY/*/$ip; do
[ -f $f ] && cat $f | while read dest; do
dip=`echo $dest | cut -d: -f1`
dport=`echo $dest | cut -d: -f2`
# send three, in case of lost packets
echo "Sending NFS tickle ack for $ip to $dip:$dport"
echo "Sending NFS tickle ack for $ip to $dest"
for i in `seq 1 3`; do
ctdb tickle $dip:$dport $ip:2049
ctdb tickle $dest $ip:2049
done
done
done
exit 0
;;
releaseip)
exit 0
;;
recovered)
exit 0
;;
shutdown)
exit 0
;;
monitor)
# always create these direcotries since NFS might be enabled at runtime
# and we dont want to restart ctdbd
mkdir -p $CTDB_BASE/state/nfstickle
mkdir -p $NFS_TICKLE_SHARED_DIRECTORY/`hostname`
mydir=$NFS_TICKLE_SHARED_DIRECTORY/`hostname`
rm -f $mydir/*
# record our connections to shared storage
netstat -tn |egrep '^tcp[[:space:]]+[0-9]+[[:space:]]+[0-9]+[[:space:]]+[0-9\.]+:2049.*ESTABLISHED' |
awk '{print $4" "$5}' |
while read dest src; do
ip=`echo $dest | cut -d: -f1`
ip=${dest%:*}
echo $src >> $mydir/$ip
done
exit 0
;;
esac

26
ctdb/config/events.d/70.iscsi
View File

@ -5,10 +5,7 @@ service_name="iscsi"
. $CTDB_BASE/functions
cmd="$1"
shift
[ "$CTDB_MANAGES_ISCSI" = "yes" ] || exit 0
ctdb_start_stop_service
[ -z "$CTDB_START_ISCSI_SCRIPTS" ] && {
echo "No iscsi start script directory found"
@ -16,15 +13,6 @@ shift
}
case $cmd in
startup)
;;
takeip)
;;
releaseip)
;;
recovered)
# block the iscsi port
iptables -I INPUT 1 -p tcp --dport 3260 -j DROP
@ -49,8 +37,8 @@ case $cmd in
done
# remove all iptables rules
while `iptables -D INPUT -p tcp --dport 3260 -j DROP 2>/dev/null >/dev/null` ; do
true;
while iptables -D INPUT -p tcp --dport 3260 -j DROP 2>/dev/null >/dev/null ; do
:
done
;;
@ -61,9 +49,11 @@ case $cmd in
;;
monitor)
[ -f $CTDB_BASE/state/iscsi/iscsi_active ] && {
ctdb_check_tcp_ports 3260 || exit $?
}
ctdb_check_tcp_ports 3260 || exit $?
;;
status)
ctdb_checkstatus || exit $?
;;
esac

16
ctdb/config/events.d/91.lvs
View File

@ -2,6 +2,7 @@
# script to manage the lvs ip multiplexer for a single public address cluster
. $CTDB_BASE/functions
loadconfig ctdb
[ -z "$CTDB_LVS_PUBLIC_IP" ] && exit 0
@ -12,12 +13,6 @@ loadconfig ctdb
exit 0
}
cmd="$1"
shift
PATH=/usr/bin:/bin:/usr/sbin:/sbin:$PATH
case $cmd in
startup)
ipvsadm -D -t $CTDB_LVS_PUBLIC_IP:0
@ -42,12 +37,6 @@ case $cmd in
echo 1 > /proc/sys/net/ipv4/route/flush
;;
takeip)
;;
releaseip)
;;
recovered|stopped)
# kill off any tcp connections
ipvsadm -D -t $CTDB_LVS_PUBLIC_IP:0
@ -89,9 +78,6 @@ case $cmd in
echo 1 > /proc/sys/net/ipv4/route/flush
;;
monitor)
;;
esac
exit 0

13
ctdb/config/events.d/99.timeout
View File

@ -9,15 +9,12 @@ loadconfig ctdb
[ "x$CTDB_RUN_TIMEOUT_MONITOR" = "xyes" ] || exit 0
cmd="$1"
shift
case $cmd in
monitor)
TIMEOUT=$(ctdb listvars | grep EventScriptTimeout | awk '{print $3}')
echo "sleeping for $((TIMEOUT * 2)) seconds..."
sleep $((TIMEOUT * 2))
;;
monitor)
TIMEOUT=$(ctdb listvars | awk '$1 == "EventScriptTimeout" {print $3}')
echo "sleeping for $((TIMEOUT * 2)) seconds..."
sleep $((TIMEOUT * 2))
;;
esac
exit 0

35
ctdb/config/functions
View File

@ -1,15 +1,20 @@
# utility functions for ctdb event scripts
PATH=/bin:/usr/bin:$PATH
PATH=/bin:/usr/bin:/usr/sbin:/sbin:$PATH
#######################################
# pull in a system config file, if any
loadconfig() {
name="$1"
if [ -n "$name" ] ; then
loadconfig ctdb
loadconfig "${service_config:-${service_name}}"
if [ "$name" != "ctdb" ] ; then
loadconfig "ctdb"
fi
if [ -z "$name" ] ; then
foo="${service_config:-${service_name}}"
if [ -n "$foo" ] ; then
loadconfig "$foo"
fi
fi
if [ -f /etc/sysconfig/$name ]; then
@ -236,7 +241,7 @@ ctdb_check_tcp_ports() {
[ $all_ok -eq 1 ] || {
echo "ERROR: $service_name tcp port $p is not responding"
exit 1
return 1
}
done
}
@ -246,8 +251,7 @@ ctdb_check_tcp_ports() {
# usage: ctdb_check_unix_socket SERVICE_NAME <socket_path>
######################################################
ctdb_check_unix_socket() {
service_name="$1"
socket_path="$2"
socket_path="$1"
[ -z "$socket_path" ] && return;
# check availability of netstat first
@ -273,7 +277,7 @@ ctdb_check_unix_socket() {
[ $all_ok -eq 1 ] || {
echo "ERROR: $service_name socket $socket_path not found"
exit 1
return 1
}
}
@ -493,15 +497,13 @@ remove_ip() {
########################################################
# some simple logic for counting events - per eventscript
# usage: ctdb_counter_init <tag>
# ctdb_counter_incr <tag>
# ctdb_counter_limit <tag> <limit>
# e.g. <tag> = "fail-count"
# ctdb_counter_limit succeeds when count >= <limit>
# usage: ctdb_counter_init
# ctdb_counter_incr
# ctdb_check_counter_limit <limit>
# ctdb_check_counter_limit succeeds when count >= <limit>
########################################################
_ctdb_counter_common () {
_eventscript="${0##*/}" # basename
_counter_file="$ctdb_fail_dir/${service_name:-${_eventscript}}"
_counter_file="$ctdb_fail_dir/$service_name"
mkdir -p "${_counter_file%/*}" # dirname
}
ctdb_counter_init () {
@ -669,5 +671,8 @@ ctdb_service_stop ()
done
}
# A reasonable default is the basename of the eventscript.
service_name="${0##*/}" # basename
ctdb_event="$1" ; shift
cmd="$ctdb_event"

4
ctdb/config/statd-callout
View File

@ -28,7 +28,7 @@ case "$1" in
add-client)
# the callout does not tell us to which ip the client connected
# so we must add it to all the ips that we serve
for f in `/bin/ls $CTDB_BASE/state/statd/ip/*`; do
for f in $CTDB_BASE/state/statd/ip/*; do
ip=`basename $f`
[ -d $STATD_SHARED_DIRECTORY/$ip ] || /bin/mkdir $STATD_SHARED_DIRECTORY/$ip
touch $STATD_SHARED_DIRECTORY/$ip/$2
@ -37,7 +37,7 @@ case "$1" in
del-client)
# the callout does not tell us to which ip the client connected
# so we must add it to all the ips that we serve
for f in `/bin/ls $CTDB_BASE/state/statd/ip/*`; do
for f in $CTDB_BASE/state/statd/ip/*; do
ip=`basename $f`
/bin/rm -f $STATD_SHARED_DIRECTORY/$ip/$2
done