sin/samba-mirror
1
0
Fork 0
mirror of https://github.com/samba-team/samba.git synced 2025-03-22 02:50:28 +03:00
Code

Merge commit 'origin/master'

Browse Source 
(This used to be ctdb commit cb00f86752d276ef0ac4dad97f7b380439603286)
This commit is contained in:
Martin Schwenke 2010-01-22 17:05:11 +11:00
commit b5254bf4ef
43 changed files with 3765 additions and 341 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
ctdb/Makefile.in
View File

@ -35,7 +35,8 @@ CFLAGS=-g -I$(srcdir)/include -Iinclude -Ilib -Ilib/util -I$(srcdir) \
LIB_FLAGS=@LDFLAGS@ -Llib @LIBS@ $(POPT_LIBS) @INFINIBAND_LIBS@ @CTDB_PCAP_LDFLAGS@
UTIL_OBJ = lib/util/idtree.o lib/util/db_wrap.o lib/util/strlist.o lib/util/util.o \
lib/util/util_time.o lib/util/util_file.o
lib/util/util_time.o lib/util/util_file.o lib/util/fault.o lib/util/substitute.o \
lib/util/signal.o
CTDB_COMMON_OBJ = common/ctdb_io.o common/ctdb_util.o \
common/ctdb_ltdb.o common/ctdb_message.o common/cmdline.o \
@ -210,6 +211,7 @@ install: all
${INSTALLCMD} -m 644 include/ctdb_private.h $(DESTDIR)$(includedir) # for samba3
${INSTALLCMD} -m 644 config/functions $(DESTDIR)$(etcdir)/ctdb
${INSTALLCMD} -m 755 config/statd-callout $(DESTDIR)$(etcdir)/ctdb
${INSTALLCMD} -m 755 config/interface_modify.sh $(DESTDIR)$(etcdir)/ctdb
${INSTALLCMD} -m 644 config/events.d/README $(DESTDIR)$(docdir)/ctdb/README.eventscripts
${INSTALLCMD} -m 644 doc/recovery-process.txt $(DESTDIR)$(docdir)/ctdb/recovery-process.txt
${INSTALLCMD} -m 755 config/events.d/00.ctdb $(DESTDIR)$(etcdir)/ctdb/events.d
@ -217,6 +219,7 @@ install: all
${INSTALLCMD} -m 755 config/events.d/10.interface $(DESTDIR)$(etcdir)/ctdb/events.d
${INSTALLCMD} -m 755 config/events.d/11.natgw $(DESTDIR)$(etcdir)/ctdb/events.d
${INSTALLCMD} -m 755 config/events.d/11.routing $(DESTDIR)$(etcdir)/ctdb/events.d
${INSTALLCMD} -m 755 config/events.d/13.per_ip_routing $(DESTDIR)$(etcdir)/ctdb/events.d
${INSTALLCMD} -m 644 config/events.d/20.multipathd $(DESTDIR)$(etcdir)/ctdb/events.d
${INSTALLCMD} -m 644 config/events.d/31.clamd $(DESTDIR)$(etcdir)/ctdb/events.d
${INSTALLCMD} -m 755 config/events.d/40.vsftpd $(DESTDIR)$(etcdir)/ctdb/events.d

176
ctdb/client/ctdb_client.c
View File

@ -2318,16 +2318,18 @@ int ctdb_ctrl_list_tunables(struct ctdb_context *ctdb,
}
int ctdb_ctrl_get_public_ips(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx, struct ctdb_all_public_ips **ips)
int ctdb_ctrl_get_public_ips_flags(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx,
uint32_t flags,
struct ctdb_all_public_ips **ips)
{
int ret;
TDB_DATA outdata;
int32_t res;
ret = ctdb_control(ctdb, destnode, 0,
CTDB_CONTROL_GET_PUBLIC_IPS, 0, tdb_null,
CTDB_CONTROL_GET_PUBLIC_IPS, flags, tdb_null,
mem_ctx, &outdata, &res, &timeout, NULL);
if (ret == 0 && res == -1) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control to get public ips failed, falling back to ipv4-only version\n"));
@ -2344,6 +2346,16 @@ int ctdb_ctrl_get_public_ips(struct ctdb_context *ctdb,
return 0;
}
int ctdb_ctrl_get_public_ips(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx,
struct ctdb_all_public_ips **ips)
{
return ctdb_ctrl_get_public_ips_flags(ctdb, timeout,
destnode, mem_ctx,
0, ips);
}
int ctdb_ctrl_get_public_ipsv4(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx, struct ctdb_all_public_ips **ips)
@ -2377,6 +2389,162 @@ int ctdb_ctrl_get_public_ipsv4(struct ctdb_context *ctdb,
return 0;
}
int ctdb_ctrl_get_public_ip_info(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx,
const ctdb_sock_addr *addr,
struct ctdb_control_public_ip_info **_info)
{
int ret;
TDB_DATA indata;
TDB_DATA outdata;
int32_t res;
struct ctdb_control_public_ip_info *info;
uint32_t len;
uint32_t i;
indata.dptr = discard_const_p(uint8_t, addr);
indata.dsize = sizeof(*addr);
ret = ctdb_control(ctdb, destnode, 0,
CTDB_CONTROL_GET_PUBLIC_IP_INFO, 0, indata,
mem_ctx, &outdata, &res, &timeout, NULL);
if (ret != 0 || res != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get public ip info "
"failed ret:%d res:%d\n",
ret, res));
return -1;
}
len = offsetof(struct ctdb_control_public_ip_info, ifaces);
if (len > outdata.dsize) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get public ip info "
"returned invalid data with size %u > %u\n",
(unsigned int)outdata.dsize,
(unsigned int)len));
dump_data(DEBUG_DEBUG, outdata.dptr, outdata.dsize);
return -1;
}
info = (struct ctdb_control_public_ip_info *)outdata.dptr;
len += info->num*sizeof(struct ctdb_control_iface_info);
if (len > outdata.dsize) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get public ip info "
"returned invalid data with size %u > %u\n",
(unsigned int)outdata.dsize,
(unsigned int)len));
dump_data(DEBUG_DEBUG, outdata.dptr, outdata.dsize);
return -1;
}
/* make sure we null terminate the returned strings */
for (i=0; i < info->num; i++) {
info->ifaces[i].name[CTDB_IFACE_SIZE] = '\0';
}
*_info = (struct ctdb_control_public_ip_info *)talloc_memdup(mem_ctx,
outdata.dptr,
outdata.dsize);
talloc_free(outdata.dptr);
if (*_info == NULL) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get public ip info "
"talloc_memdup size %u failed\n",
(unsigned int)outdata.dsize));
return -1;
}
return 0;
}
int ctdb_ctrl_get_ifaces(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx,
struct ctdb_control_get_ifaces **_ifaces)
{
int ret;
TDB_DATA outdata;
int32_t res;
struct ctdb_control_get_ifaces *ifaces;
uint32_t len;
uint32_t i;
ret = ctdb_control(ctdb, destnode, 0,
CTDB_CONTROL_GET_IFACES, 0, tdb_null,
mem_ctx, &outdata, &res, &timeout, NULL);
if (ret != 0 || res != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get ifaces "
"failed ret:%d res:%d\n",
ret, res));
return -1;
}
len = offsetof(struct ctdb_control_get_ifaces, ifaces);
if (len > outdata.dsize) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get ifaces "
"returned invalid data with size %u > %u\n",
(unsigned int)outdata.dsize,
(unsigned int)len));
dump_data(DEBUG_DEBUG, outdata.dptr, outdata.dsize);
return -1;
}
ifaces = (struct ctdb_control_get_ifaces *)outdata.dptr;
len += ifaces->num*sizeof(struct ctdb_control_iface_info);
if (len > outdata.dsize) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get ifaces "
"returned invalid data with size %u > %u\n",
(unsigned int)outdata.dsize,
(unsigned int)len));
dump_data(DEBUG_DEBUG, outdata.dptr, outdata.dsize);
return -1;
}
/* make sure we null terminate the returned strings */
for (i=0; i < ifaces->num; i++) {
ifaces->ifaces[i].name[CTDB_IFACE_SIZE] = '\0';
}
*_ifaces = (struct ctdb_control_get_ifaces *)talloc_memdup(mem_ctx,
outdata.dptr,
outdata.dsize);
talloc_free(outdata.dptr);
if (*_ifaces == NULL) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control for get ifaces "
"talloc_memdup size %u failed\n",
(unsigned int)outdata.dsize));
return -1;
}
return 0;
}
int ctdb_ctrl_set_iface_link(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx,
const struct ctdb_control_iface_info *info)
{
int ret;
TDB_DATA indata;
int32_t res;
indata.dptr = discard_const_p(uint8_t, info);
indata.dsize = sizeof(*info);
ret = ctdb_control(ctdb, destnode, 0,
CTDB_CONTROL_SET_IFACE_LINK_STATE, 0, indata,
mem_ctx, NULL, &res, &timeout, NULL);
if (ret != 0 || res != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_control for set iface link "
"failed ret:%d res:%d\n",
ret, res));
return -1;
}
return 0;
}
/*
set/clear the permanent disabled bit on a remote node
*/

24
ctdb/common/ctdb_util.c
View File

@ -367,7 +367,7 @@ bool parse_ipv4(const char *s, unsigned port, struct sockaddr_in *sin)
return true;
}
static bool parse_ipv6(const char *s, const char *iface, unsigned port, ctdb_sock_addr *saddr)
static bool parse_ipv6(const char *s, const char *ifaces, unsigned port, ctdb_sock_addr *saddr)
{
saddr->ip6.sin6_family = AF_INET6;
saddr->ip6.sin6_port = htons(port);
@ -379,8 +379,14 @@ static bool parse_ipv6(const char *s, const char *iface, unsigned port, ctdb_soc
return false;
}
if (iface && IN6_IS_ADDR_LINKLOCAL(&saddr->ip6.sin6_addr)) {
saddr->ip6.sin6_scope_id = if_nametoindex(iface);
if (ifaces && IN6_IS_ADDR_LINKLOCAL(&saddr->ip6.sin6_addr)) {
if (strchr(ifaces, ',')) {
DEBUG(DEBUG_ERR, (__location__ " Link local address %s "
"is specified for multiple ifaces %s\n",
s, ifaces));
return false;
}
saddr->ip6.sin6_scope_id = if_nametoindex(ifaces);
}
return true;
@ -430,7 +436,7 @@ bool parse_ip_port(const char *addr, ctdb_sock_addr *saddr)
/*
parse an ip
*/
bool parse_ip(const char *addr, const char *iface, unsigned port, ctdb_sock_addr *saddr)
bool parse_ip(const char *addr, const char *ifaces, unsigned port, ctdb_sock_addr *saddr)
{
char *p;
bool ret;
@ -440,7 +446,7 @@ bool parse_ip(const char *addr, const char *iface, unsigned port, ctdb_sock_addr
if (p == NULL) {
ret = parse_ipv4(addr, port, &saddr->ip);
} else {
ret = parse_ipv6(addr, iface, port, saddr);
ret = parse_ipv6(addr, ifaces, port, saddr);
}
return ret;
@ -449,7 +455,7 @@ bool parse_ip(const char *addr, const char *iface, unsigned port, ctdb_sock_addr
/*
parse a ip/mask pair
*/
bool parse_ip_mask(const char *str, const char *iface, ctdb_sock_addr *addr, unsigned *mask)
bool parse_ip_mask(const char *str, const char *ifaces, ctdb_sock_addr *addr, unsigned *mask)
{
TALLOC_CTX *tmp_ctx = talloc_new(NULL);
char *s, *p;
@ -482,7 +488,7 @@ bool parse_ip_mask(const char *str, const char *iface, ctdb_sock_addr *addr, uns
/* now is this a ipv4 or ipv6 address ?*/
ret = parse_ip(s, iface, 0, addr);
ret = parse_ip(s, ifaces, 0, addr);
talloc_free(tmp_ctx);
return ret;
@ -653,6 +659,7 @@ void ctdb_lockdown_memory(struct ctdb_context *ctdb)
}
const char *ctdb_eventscript_call_names[] = {
"init",
"startup",
"startrecovery",
"recovered",
@ -662,5 +669,6 @@ const char *ctdb_eventscript_call_names[] = {
"monitor",
"status",
"shutdown",
"reload"
"reload",
"updateip"
};

59
ctdb/config/ctdb.sysconfig
View File

@ -179,6 +179,65 @@ CTDB_RECOVERY_LOCK="/some/place/on/shared/storage"
# CTDB_NATGW_PRIVATE_NETWORK=10.1.1.0/24
# CTDB_NATGW_NODES=/etc/ctdb/natgw_nodes
# PER_IP_ROUTING configuration
#
# Some setups have multiple network interfaces connected to the
# same network. By default all traffic for a network is routed
# through only one interface, while the others are idle.
#
# On Linux it possible to use policy based routing to spread the load
# across all interfaces. The is implemented by using a separate
# routing table per public ip address.
#
# The configuration file configured by CTDB_PER_IP_ROUTING_CONF
# contains the list of additional routes. The routes are bound to the
# interface that is holding the public ip address.
#
# The format of the config file looks like this:
# <public_ip_address> <network> [<gateway>]
# and it's possible to have multiple routes per public ip address.
#
# If the special value "__auto_link_local__" is used, the config
# file autogenerated. Each public ip address gets a special route
# for its own subnet bound to it's current interface.
# E.g. 10.1.2.3/24 will result in a config file line
# 10.1.2.3 10.1.2.0/24
#
# The CTDB_PER_IP_ROUTING_RULE_PREF option needs to be configured.
# The value will be passed as "pref" argument of "ip rule".
# The value should be between 1 and 32765. So that the rule
# comes after the rule for "local" routing table and before
# the rule for the "main" routing table. This way the specific
# routing table just overloads the "main" routing table,
# this is useful because with the "__auto_link_local__" setup
# the default route still comes from the "main" routing table.
#
# The routing table ids are automaticly allocated. On
# Linux the routing table ids must be in the range of 0 to 255.
# But some are reserved values, see /etc/iproute2/rt_tables.
# You need to configure a range (CTDB_PER_IP_ROUTING_TABLE_ID_LOW
# and CTDB_PER_IP_ROUTING_TABLE_ID_HIGH) from which the table ids can be taken.
#
# The default value for CTDB_PER_IP_ROUTING_CONF is "",
# which means the feature is disabled by default.
#
# CTDB_PER_IP_ROUTING_CONF="/etc/ctdb/per_ip_routing.conf"
# CTDB_PER_IP_ROUTING_CONF="__auto_link_local__"
# CTDB_PER_IP_ROUTING_TABLE_ID_LOW=10
# CTDB_PER_IP_ROUTING_TABLE_ID_HIGH=250
# CTDB_PER_IP_ROUTING_RULE_PREF=10000
# Make offline interfaces not a reason for being UNHEALTHY.
# The CTDB_PARTIALLY_ONLINE_INTERFACES option changes
# the behavior of the 10.interface monitor event.
# In some setups it's desired that interfaces without
# an active link don't change the node to unhealthy.
# ctdbd is just informed about the interface status
# and "ctdb status" dislays the node as "PARTIALLYONLINE".
#
# CTDB_PARTIALLY_ONLINE_INTERFACES="yes"
# where to log messages
# the default is /var/log/log.ctdb
# CTDB_LOGFILE=/var/log/log.ctdb

2
ctdb/config/events.d/00.ctdb
View File

@ -13,7 +13,7 @@
loadconfig
case "$1" in
startup)
init)
# make sure we have a blank state directory for the scripts to work with
/bin/rm -rf $CTDB_BASE/state
/bin/mkdir -p $CTDB_BASE/state

2
ctdb/config/events.d/01.reclock
View File

@ -5,7 +5,7 @@
loadconfig
case "$1" in
startup)
init)
ctdb_counter_init
;;

261
ctdb/config/events.d/10.interface
View File

@ -17,10 +17,114 @@ loadconfig
exit 0
}
monitor_interfaces()
{
local INTERFACES=`cat $CTDB_PUBLIC_ADDRESSES |
sed -e "s/^[^\t ]*[\t ]*//" -e "s/,/ /g" -e "s/[\t ]*$//"`
[ "$CTDB_PUBLIC_INTERFACE" ] && INTERFACES="$CTDB_PUBLIC_INTERFACE $INTERFACES"
[ "$CTDB_NATGW_PUBLIC_IFACE" ] && INTERFACES="$CTDB_NATGW_PUBLIC_IFACE $INTERFACES"
local IFACES=`ctdb ifaces -Y | grep -v '^:Name:LinkStatus:References:'`
local I
local IFACE
for I in $IFACES; do
IFACE=`echo -n "$I" | cut -d ':' -f2`
INTERFACES="$IFACE $INTERFACES"
done
INTERFACES=`for IFACE in $INTERFACES ; do echo $IFACE ; done | sort | uniq`
local fail=0
local force_fail=0
local ok=0
for IFACE in $INTERFACES ; do
local OLDLINK=`echo -n "$IFACES" | grep "^:$IFACE:" | cut -d ':' -f3 | xargs`
test -z "$OLDLINK" && {
force_fail=1
}
# These interfaces are sometimes bond devices
# When we use VLANs for bond interfaces, there will only
# be an entry in /proc for the underlying real interface
local REALIFACE=`echo $IFACE |sed -e 's/\..*$//'`
[ -f /proc/net/bonding/$REALIFACE ] && {
grep -q 'Currently Active Slave: None' /proc/net/bonding/$REALIFACE && {
echo "ERROR: No active slaves for bond device $REALIFACE"
fail=1
test -n "$OLDLINK" && {
ctdb setifacelink $IFACE down
}
continue;
}
grep -q '^MII Status: up' /proc/net/bonding/$REALIFACE || {
echo "ERROR: public network interface $REALIFACE is down"
fail=1
test -n "$OLDLINK" && {
ctdb setifacelink $IFACE down
}
continue;
}
test -n "$OLDLINK" && {
ok=1 # we only set ok for interfaces known to ctdbd
ctdb setifacelink $IFACE up
}
return 0;
}
case $IFACE in
ib*)
# we dont know how to test ib links
;;
*)
[ -z "$IFACE" ] || {
ethtool $IFACE | grep -q 'Link detected: yes' || {
# On some systems, this is not successful when a
# cable is plugged but the interface has not been
# brought up previously. Bring the interface up and
# try again...
/sbin/ip link set $IFACE up
ethtool $IFACE | grep -q 'Link detected: yes' || {
echo "ERROR: No link on the public network interface $IFACE"
fail=1
test -n "$OLDLINK" && {
ctdb setifacelink $IFACE down
}
continue
}
}
test -n "$OLDLINK" && {
ok=1 # we only set ok for interfaces known to ctdbd
ctdb setifacelink $IFACE up
}
}
;;
esac
done
test x"$fail" = x"0" && {
return 0;
}
test x"$force_fail" != x"0" && {
return 1;
}
test x"$ok" = x"1" && {
return 2;
}
return 1;
}
case "$1" in
#############################
# called when ctdbd starts up
startup)
init)
# make sure that we only respond to ARP messages from the NIC where
# a particular ip address is associated.
[ -f /proc/sys/net/ipv4/conf/all/arp_filter ] && {
@ -37,6 +141,14 @@ case "$1" in
done
;;
#############################
# called after ctdbd has done its initial recovery
# and we start the services to become healthy
startup)
monitor_interfaces
;;
################################################
# called when ctdbd wants to claim an IP address
@ -49,14 +161,10 @@ case "$1" in
ip=$3
maskbits=$4
# we make sure the interface is up first
/sbin/ip link set $iface up || {
echo "Failed to bringup interface $iface"
exit 1
}
/sbin/ip addr add $ip/$maskbits brd + dev $iface || {
echo "Failed to add $ip/$maskbits on dev $iface"
add_ip_to_iface $iface $ip $maskbits || {
exit 1;
}
# cope with the script being killed while we have the interface blocked
iptables -D INPUT -i $iface -d $ip -j DROP 2> /dev/null
@ -93,34 +201,74 @@ case "$1" in
iptables -I INPUT -i $iface -d $ip -j DROP
kill_tcp_connections $ip
# the ip tool will delete all secondary IPs if this is the primary. To work around
# this _very_ annoying behaviour we have to keep a record of the secondaries and re-add
# them afterwards. yuck
secondaries=""
if /sbin/ip addr list dev $iface primary | grep -q "inet $ip/$maskbits " ; then
secondaries=`/sbin/ip addr list dev $iface secondary | grep " inet " | awk '{print $2}'`
fi
/sbin/ip addr del $ip/$maskbits dev $iface || failed=1
[ -z "$secondaries" ] || {
for i in $secondaries; do
if /sbin/ip addr list dev $iface | grep -q "inet $i" ; then
echo "kept secondary $i on dev $iface"
else
echo "re-adding secondary address $i to dev $iface"
/sbin/ip addr add $i dev $iface || failed=1
fi
done
delete_ip_from_iface $iface $ip $maskbits || {
iptables -D INPUT -i $iface -d $ip -j DROP 2> /dev/null
exit 1;
}
iptables -D INPUT -i $iface -d $ip -j DROP 2> /dev/null
[ $failed = 0 ] || {
echo "Failed to del $ip on dev $iface"
exit 1
}
# flush our route cache
echo 1 > /proc/sys/net/ipv4/route/flush
;;
##################################################
# called when ctdbd wants to update an IP address
updateip)
if [ $# != 5 ]; then
echo "must supply old interface, new interface, IP and maskbits"
exit 1
fi
# moving an IP is a bit more complex than it seems.
# First we drop all traffic on the old interface.
# Then we try to add the ip to the new interface and before
# we finally remove it from the old interface.
#
# 1) firewall this IP, so no new external packets arrive for it
# 2) add the IP to the new interface
# 3) remove the IP from the old interface
# 4) remove the firewall rule
# 5) use ctdb gratiousarp to propagate the new mac address
# 6) use netstat -tn to find existing connections, and tickle them
oiface=$2
niface=$3
ip=$4
maskbits=$5
failed=0
# we do an extra delete to cope with the script being killed
iptables -D INPUT -i $oiface -d $ip -j DROP 2> /dev/null
iptables -I INPUT -i $oiface -d $ip -j DROP
# we make sure the interface is up first
add_ip_to_iface $niface $ip $maskbits || {
iptables -D INPUT -i $oiface -d $ip -j DROP 2> /dev/null
exit 1;
}
delete_ip_from_iface $oiface $ip $maskbits || {
delete_ip_from_iface $niface $ip $maskbits
iptables -D INPUT -i $oiface -d $ip -j DROP 2> /dev/null
exit 1;
}
# cope with the script being killed while we have the interface blocked
iptables -D INPUT -i $oiface -d $ip -j DROP 2> /dev/null
# flush our route cache
echo 1 > /proc/sys/net/ipv4/route/flush
# propagate the new mac address
ctdb gratiousarp $ip $niface
# tickle all existing connections, so that dropped packets
# are retransmited and the tcp streams work
tickle_tcp_connections $ip
;;
###########################################
# called when ctdbd has finished a recovery
@ -133,52 +281,21 @@ case "$1" in
;;
monitor)
INTERFACES=`cat $CTDB_PUBLIC_ADDRESSES |
sed -e "s/^[^\t ]*[\t ]*//" -e "s/[\t ]*$//"`
monitor_interfaces
ret=$?
[ "$CTDB_PUBLIC_INTERFACE" ] && INTERFACES="$CTDB_PUBLIC_INTERFACE $INTERFACES"
[ "$CTDB_NATGW_PUBLIC_IFACE" ] && INTERFACES="$CTDB_NATGW_PUBLIC_IFACE $INTERFACES"
INTERFACES=`for IFACE in $INTERFACES ; do echo $IFACE ; done | sort | uniq`
for IFACE in $INTERFACES ; do
# These interfaces are sometimes bond devices
# When we use VLANs for bond interfaces, there will only
# be an entry in /proc for the underlying real interface
REALIFACE=`echo $IFACE |sed -e 's/\..*$//'`
[ -f /proc/net/bonding/$REALIFACE ] && {
grep -q 'Currently Active Slave: None' /proc/net/bonding/$REALIFACE && {
echo "ERROR: No active slaves for bond device $REALIFACE"
exit 1
test x"$ret" = x"2" && {
test x"$CTDB_PARTIALLY_ONLINE_INTERFACES" != x"yes" && {
exit 1;
}
grep -q '^MII Status: up' /proc/net/bonding/$REALIFACE || {
echo "ERROR: public network interface $REALIFACE is down"
exit 1
}
exit 0;
}
# as long as we have one interface available don't become
# unhealthy
ret=0
}
case $IFACE in
ib*)
# we dont know how to test ib links
;;
*)
[ -z "$IFACE" ] || {
/usr/sbin/ethtool $IFACE | grep -q 'Link detected: yes' || {
# On some systems, this is not successful when a
# cable is plugged but the interface has not been
# brought up previously. Bring the interface up and
# try again...
/sbin/ip link set $IFACE up
/usr/sbin/ethtool $IFACE | grep -q 'Link detected: yes' || {
echo "ERROR: No link on the public network interface $IFACE"
exit 1
}
}
}
;;
esac
done
test x"$ret" != x"0" && {
exit 1;
}
;;
*)
ctdb_standard_event_handler "$@"

444
ctdb/config/events.d/13.per_ip_routing Executable file
View File

@ -0,0 +1,444 @@
#!/bin/sh
. $CTDB_BASE/functions
loadconfig
[ -z "$CTDB_PER_IP_ROUTING_STATE" ] && {
CTDB_PER_IP_ROUTING_STATE="$CTDB_BASE/state/per_ip_routing"
}
AUTO_LINK_LOCAL="no"
case "$CTDB_PER_IP_ROUTING_CONF" in
__auto_link_local__)
AUTO_LINK_LOCAL="yes"
CTDB_PER_IP_ROUTING_CONF="$CTDB_PER_IP_ROUTING_STATE/auto_link_local.conf"
;;
*)
[ -z "$CTDB_PER_IP_ROUTING_CONF" ] && {
#echo "No config file found. Nothing to do for 13.per_ip_routing"
exit 0;
}
;;
esac
_low=$CTDB_PER_IP_ROUTING_TABLE_ID_LOW
_high=$CTDB_PER_IP_ROUTING_TABLE_ID_HIGH
test -z "$_low" && {
echo "$0: CTDB_PER_IP_ROUTING_TABLE_ID_LOW not configured";
exit 1;
}
test -z "$_high" && {
echo "$0: CTDB_PER_IP_ROUTING_TABLE_ID_HIGH not configured";
exit 1;
}
test "$_low" -ge "$_high" && {
echo "$0: CTDB_PER_IP_ROUTING_TABLE_ID_LOW[$_low] needs to be below CTDB_PER_IP_ROUTING_TABLE_ID_HIGH[$_high]";
exit 1;
}
test -z "$CTDB_PER_IP_ROUTING_RULE_PREF" && {
echo "$0: CTDB_PER_IP_ROUTING_RULE_PREF not configured";
exit 1;
}
locknesting=0
lock_root="$CTDB_PER_IP_ROUTING_STATE"
host=`hostname`
lock_debug()
{
echo -n ""
}
############################
# grab a lock file. Not atomic, but close :)
# tries to cope with NFS
lock_file() {
if [ -z "$lock_root" ]; then
lock_root=`pwd`;
fi
lckf="$lock_root/$1"
machine=`cat "$lckf" 2> /dev/null | cut -d: -f1`
pid=`cat "$lckf" 2> /dev/null | cut -d: -f2`
if [ "$pid" = "$$" ]; then
locknesting=`expr $locknesting + 1`
lock_debug "lock nesting now $locknesting"
return 0
fi
if test -f "$lckf"; then
test $machine = $host || {
lock_debug "lock file $lckf is valid for other machine $machine"
stat -c%y "$lckf"
return 1
}
kill -0 $pid && {
lock_debug "lock file $lckf is valid for process $pid"
stat -c%y "$lckf"
return 1
}
lock_debug "stale lock file $lckf for $machine:$pid"
cat "$lckf"
/bin/rm -f "$lckf"
fi
echo "$host:$$" > "$lckf"
return 0
}
############################
# unlock a lock file
unlock_file() {
if [ -z "$lock_root" ]; then
lock_root=`pwd`;
fi
if [ "$locknesting" != "0" ]; then
locknesting=`expr $locknesting - 1`
lock_debug "lock nesting now $locknesting"
else
lckf="$lock_root/$1"
/bin/rm -f "$lckf"
fi
}
generate_table_id () {
local _ip=$1
local _ipsdir="$CTDB_PER_IP_ROUTING_STATE/ips"
local _ipdir="$_ipsdir/$_ip"
mkdir -p $_ipdir
#echo "generate_table_id $_ip"
local _id=`cat $_ipdir/table_id 2>/dev/null| xargs`
test -n "$_id" && {
#echo "IP: $_ip => OLD TABLE: $_id"
table_id=$_id
return 0;
}
local _low="$CTDB_PER_IP_ROUTING_TABLE_ID_LOW"
local _high="$CTDB_PER_IP_ROUTING_TABLE_ID_HIGH"
local _newid=""
for _id in `seq $_low $_high | xargs`; do
local _table_lck="table_id_$_id.lock"
lock_file $_table_lck 2>/dev/null || {
continue;
}
local _taken=`grep "^$_id$" $_ipsdir/*/table_id 2>/dev/null| wc -l | xargs`
test x"$_taken" != x"0" && {
unlock_file $_table_lck
#echo "tableid: $_id taken"
continue
}
_newid=$_id;
echo "$_newid" > $_ipdir/table_id
unlock_file $_table_lck
break;
done
test -z "$_newid" && {
echo "generate_table_id: out of table ids: $_low - $_high"
exit 1;
}
#echo "IP: $_ip => NEW TABLE: $_newid"
table_id=$_newid
return 0;
}
run_release_script_once()
{
local _script=$1
#echo "run_release_script_once[$_script]"
test -x "$_script" && {
#echo "run it: start"
$_script || {
echo "release_script: $_script - failed $?"
return $?;
}
#echo "run it: end"
}
echo -e "#!/bin/sh\n#\n" > $_script
chmod +x $_script
return 0;
}
generate_auto_link_local()
{
local _ip=$1
local _maskbits=$2
#echo "generate_auto_link_local $_ip $_maskbits"
local _netip=`ipv4_host_addr_to_net_addr $_ip $_maskbits`
local _line="$_ip $_netip/$_maskbits"
local _config=`cat $CTDB_PER_IP_ROUTING_CONF 2>/dev/null`
local _exact=`echo -n "$_config" | grep "^$line$" | wc -l | xargs`
test x"$_exact" = x"1" && {
return 0;
}
local _tmp="$CTDB_PER_IP_ROUTING_CONF.$$.tmp"
echo -n "$_config" | grep -v "^$_ip " > $_tmp
echo "$_line" >> $_tmp
mv $_tmp $CTDB_PER_IP_ROUTING_CONF
}
generate_per_ip_routing()
{
local _ip=$1
local _maskbits=$2
local _iface=$3
local _readonly=$4
local _ipdir="$CTDB_PER_IP_ROUTING_STATE/ips/$_ip"
table_id=""
release_script="$_ipdir/release_script.sh"
test x"$_readonly" = x"yes" && {
test -d $_ipdir || {
return 1;
}
return 0;
}
mkdir -p $_ipdir || {
echo "mkdir -p $_ipdir failed"
return 1;
}
echo "$_ip" > $_ipdir/ip
generate_table_id $_ip
test x"$AUTO_LINK_LOCAL" = x"yes" && {
generate_auto_link_local $_ip $_maskbits
}
release_script="$_ipdir/release_script.sh"
run_release_script_once $release_script
return 0;
}
case "$1" in
#############################
# called when ctdbd starts up
startup)
# cleanup old rules
pref=$CTDB_PER_IP_ROUTING_RULE_PREF
rules=`ip rule show | grep "^$pref:" | sed -e 's/.*from \([^ ][^ ]*\) lookup \([^ ][^ ]*\)/\2;\1/' | xargs`
for r in $rules; do
table_id=`echo -n "$r" | cut -d ';' -f1`
ip=`echo -n "$r" | cut -d ';' -f2-`
echo "Removing ip rule for public address $ip for routing table $table_id"
cmd="ip rule del from $ip table $table_id pref $pref"
#echo $cmd
eval $cmd
cmd="ip route flush table $table_id"
#echo $cmd
eval $cmd 2>/dev/null
done
# make sure that we only respond to ARP messages from the NIC where
# a particular ip address is associated.
[ -f /proc/sys/net/ipv4/conf/all/arp_filter ] && {
echo 1 > /proc/sys/net/ipv4/conf/all/arp_filter
}
mkdir -p $CTDB_PER_IP_ROUTING_STATE
;;
shutdown)
for s in $CTDB_PER_IP_ROUTING_STATE/ips/*/release_script.sh; do
run_release_script_once "$s"
done
rm -rf $CTDB_PER_IP_ROUTING_STATE
;;
################################################
# called when ctdbd wants to claim an IP address
takeip)
if [ $# != 4 ]; then
echo "must supply interface, IP and maskbits"
exit 1
fi
iface=$2
ip=$3
maskbits=$4
ipv4_is_valid_addr $ip || {
echo "$0: $1 not an ipv4 address skipping IP:$ip"
exit 0;
}
[ ! -d "$CTDB_PER_IP_ROUTING_STATE" ] && {
echo "$0: $1 No state directory found, waiting for startup."
exit 0;
}
generate_per_ip_routing $ip $maskbits $iface "no" || {
echo "$0: $1: generate_per_ip_routing $ip $maskbits $iface no - failed"
exit 1;
}
config=`cat $CTDB_PER_IP_ROUTING_CONF`
lines=`echo -n "$config" | grep -n "^$ip " | cut -d ':' -f1 | xargs`
pref="$CTDB_PER_IP_ROUTING_RULE_PREF"
test -n "$lines" && {
echo "ip rule del from $ip pref $pref table $table_id" >> $release_script
echo "ip route flush table $table_id 2>/dev/null" >> $release_script
ip rule del from $ip pref $pref 2>/dev/null
ip rule add from $ip pref $pref table $table_id
ip route flush table $table_id 2>/dev/null
}
for l in $lines; do
line=`echo -n "$config" | head -n $l | tail -n 1`
dest=`echo -n "$line" | cut -d ' ' -f 2`
gw=`echo -n "$line" | cut -d ' ' -f 3`
via=""
test -n "$gw" && {
via="via $gw"
}
ip route add $dest $via dev $iface table $table_id
done
# flush our route cache
echo 1 > /proc/sys/net/ipv4/route/flush
ctdb gratiousarp $ip $iface
;;
################################################
# called when ctdbd wants to claim an IP address
updateip)
if [ $# != 5 ]; then
echo "must supply old interface, new interface, IP and maskbits"
exit 1
fi
oiface=$2
niface=$3
ip=$4
maskbits=$5
ipv4_is_valid_addr $ip || {
echo "$0: $1 not an ipv4 address skipping IP:$ip"
exit 0;
}
[ ! -d "$CTDB_PER_IP_ROUTING_STATE" ] && {
echo "$0: $1 No state directory found, waiting for startup."
exit 0;
}
generate_per_ip_routing $ip $maskbits $niface "no" || {
echo "$0: $1: generate_per_ip_routing $ip $maskbits $niface no - failed"
exit 1;
}
config=`cat $CTDB_PER_IP_ROUTING_CONF`
lines=`echo -n "$config" | grep -n "^$ip " | cut -d ':' -f1 | xargs`
pref="$CTDB_PER_IP_ROUTING_RULE_PREF"
test -n "$lines" && {
echo "ip rule del from $ip pref $pref table $table_id" >> $release_script
echo "ip route flush table $table_id 2>/dev/null" >> $release_script
ip rule del from $ip pref $pref 2>/dev/null
ip rule add from $ip pref $pref table $table_id
ip route flush table $table_id 2>/dev/null
}
for l in $lines; do
line=`echo -n "$config" | head -n $l | tail -n 1`
dest=`echo -n "$line" | cut -d ' ' -f 2`
gw=`echo -n "$line" | cut -d ' ' -f 3`
via=""
test -n "$gw" && {
via="via $gw"
}
ip route add $dest $via dev $niface table $table_id
done
# flush our route cache
echo 1 > /proc/sys/net/ipv4/route/flush
ctdb gratiousarp $ip $niface
tickle_tcp_connections $ip
;;
##################################################
# called when ctdbd wants to release an IP address
releaseip)
if [ $# != 4 ]; then
echo "must supply interface, IP and maskbits"
exit 1
fi
iface=$2
ip=$3
maskbits=$4
ipv4_is_valid_addr $ip || {
echo "$0: $1 not an ipv4 address skipping IP:$ip"
exit 0;
}
[ ! -d "$CTDB_PER_IP_ROUTING_STATE" ] && {
echo "$0: $1 No state directory found, waiting for startup."
exit 0;
}
generate_per_ip_routing $ip $maskbits $iface "yes" || {
echo "$0: $1: generate_per_ip_routing $ip $maskbits $iface yes - failed"
exit 1;
}
run_release_script_once "$release_script"
;;
###########################################
# called when ctdbd has finished a recovery
recovered)
;;
####################################
# called when ctdbd is shutting down
shutdown)
;;
monitor)
;;
*)
ctdb_standard_event_handler "$@"
;;
esac
exit 0

23
ctdb/config/events.d/README
View File

@ -22,11 +22,20 @@ All of the events except the 'shutdown' and 'startrecovery' events will be
called with the ctdb daemon in NORMAL mode (ie. not in recovery)
The events currently implemented are
startup
init
This event does not take any additional arguments.
This event is only invoked once, when ctdb is starting up.
This event is used to wait for the service to start and all
resources for the service becoming available.
This event is used to do some cleanup work from earlier runs
and prepare the basic setup.
Example: 00.ctdb cleans up $CTDB_BASE/state
startup
This event does not take any additional arguments.
This event is only invoked once, when ctdb has finished
the initial recoveries. This event is used to wait for
the service to start and all resources for the service
becoming available.
This is used to prevent ctdb from starting up and advertize its
services until all dependent services have become available.
@ -105,6 +114,14 @@ releaseip
Example: 60.nfs
updateip
This event is triggered everytime the node moves a public ip
address between interfaces
This event takes four additional arguments :
'old-interface' 'new-interface' 'ipaddress' and 'netmask'
Example: 10.interface
startrecovery
This event is triggered everytime we start a recovery process
or before we start changing ip address allocations.

128
ctdb/config/functions
View File

@ -307,6 +307,36 @@ kill_tcp_connections_local_only() {
echo "killed $_killcount TCP connections to released IP $_IP"
}
##################################################################
# tickle any TCP connections with the given IP
##################################################################
tickle_tcp_connections() {
_IP="$1"
_failed=0
_killcount=0
connfile="$CTDB_BASE/state/connections.$_IP"
netstat -tn |egrep "^tcp.*[[:space:]]+$_IP:.*ESTABLISHED" | awk '{print $4" "$5}' > $connfile
netstat -tn |egrep "^tcp.*[[:space:]]+::ffff:$_IP:.*ESTABLISHED" | awk '{print $4" "$5}' >> $connfile
while read dest src; do
srcip=`echo $src | sed -e "s/:[^:]*$//"`
srcport=`echo $src | sed -e "s/^.*://"`
destip=`echo $dest | sed -e "s/:[^:]*$//"`
destport=`echo $dest | sed -e "s/^.*://"`
echo "Tickle TCP connection $srcip:$srcport $destip:$destport"
ctdb tickle $srcip:$srcport $destip:$destport >/dev/null 2>&1 || _failed=1
echo "Tickle TCP connection $destip:$destport $srcip:$srcport"
ctdb tickle $destip:$destport $srcip:$srcport >/dev/null 2>&1 || _failed=1
done < $connfile
/bin/rm -f $connfile
[ $_failed = 0 ] || {
echo "Failed to send tickle control"
return;
}
}
########################################################
# start/stop the nfs service on different platforms
########################################################
@ -391,9 +421,7 @@ startstop_nfslock() {
esac
}
########################################################
# remove an ip address from an interface
########################################################
# better use delete_ip_from_iface() together with add_ip_to_iface
remove_ip() {
# the ip tool will delete all secondary IPs if this is the primary.
# To work around this _very_ annoying behaviour we have to keep a
@ -415,6 +443,36 @@ remove_ip() {
}
}
add_ip_to_iface()
{
local _iface=$1
local _ip=$2
local _maskbits=$3
local _lockfile="$CTDB_BASE/state/interface_modify.$_iface.flock"
test -f $_lockfile || {
touch $_lockfile
}
flock --timeout 30 $_lockfile $CTDB_BASE/interface_modify.sh add "$_iface" "$_ip" "$_maskbits"
return $?
}
delete_ip_from_iface()
{
local _iface=$1
local _ip=$2
local _maskbits=$3
local _lockfile="$CTDB_BASE/state/interface_modify.$_iface.flock"
test -f $_lockfile || {
touch $_lockfile
}
flock --timeout 30 $_lockfile $CTDB_BASE/interface_modify.sh delete "$_iface" "$_ip" "$_maskbits"
return $?
}
########################################################
# some simple logic for counting events - per eventscript
# usage: ctdb_counter_init
@ -600,6 +658,70 @@ ctdb_standard_event_handler ()
esac
}
ipv4_host_addr_to_net_addr()
{
local HOST=$1
local MASKBITS=$2
local HOST0=$(echo $HOST | awk -F . '{print $4}')
local HOST1=$(echo $HOST | awk -F . '{print $3}')
local HOST2=$(echo $HOST | awk -F . '{print $2}')
local HOST3=$(echo $HOST | awk -F . '{print $1}')
local HOST_NUM=$(( $HOST0 + $HOST1 * 256 + $HOST2 * (256 ** 2) + $HOST3 * (256 ** 3) ))
local MASK_NUM=$(( ( (2**32 - 1) * (2**(32 - $MASKBITS)) ) & (2**32 - 1) ))
local NET_NUM=$(( $HOST_NUM & $MASK_NUM))
local NET0=$(( $NET_NUM & 255 ))
local NET1=$(( ($NET_NUM & (255 * 256)) / 256 ))
local NET2=$(( ($NET_NUM & (255 * 256**2)) / 256**2 ))
local NET3=$(( ($NET_NUM & (255 * 256**3)) / 256**3 ))
echo "$NET3.$NET2.$NET1.$NET0"
}
ipv4_maskbits_to_net_mask()
{
local MASKBITS=$1
local MASK_NUM=$(( ( (2**32 - 1) * (2**(32 - $MASKBITS)) ) & (2**32 - 1) ))
local MASK0=$(( $MASK_NUM & 255 ))
local MASK1=$(( ($MASK_NUM & (255 * 256)) / 256 ))
local MASK2=$(( ($MASK_NUM & (255 * 256**2)) / 256**2 ))
local MASK3=$(( ($MASK_NUM & (255 * 256**3)) / 256**3 ))
echo "$MASK3.$MASK2.$MASK1.$MASK0"
}
ipv4_is_valid_addr()
{
local ADDR=$1
local fail=0
local N=`echo $ADDR | sed -e 's/[0-9][0-9]*\.[0-9][0-9]*\.[0-9][0-9]*\.[0-9][0-9]*//'`
test -n "$N" && fail=1
local ADDR0=$(echo $ADDR | awk -F . '{print $4}')
local ADDR1=$(echo $ADDR | awk -F . '{print $3}')
local ADDR2=$(echo $ADDR | awk -F . '{print $2}')
local ADDR3=$(echo $ADDR | awk -F . '{print $1}')
test "$ADDR0" -gt 255 && fail=1
test "$ADDR1" -gt 255 && fail=1
test "$ADDR2" -gt 255 && fail=1
test "$ADDR3" -gt 255 && fail=1
test x"$fail" != x"0" && {
#echo "IPv4: '$ADDR' is not a valid address"
return 1;
}
return 0;
}
########################################################
# load a site local config file
########################################################

87
ctdb/config/gdb_backtrace Executable file
View File

@ -0,0 +1,87 @@
#!/bin/sh
BASENAME=`basename $0`
if [ -n "$VALGRIND" -o -n "$SMBD_VALGRIND" ]; then
echo "${BASENAME}: Not running debugger under valgrind"
exit 1
fi
# we want everything on stderr, so the program is not disturbed
exec 1>&2
BASENAME=`basename $0`
UNAME=`uname`
PID=$1
BINARY=$2
test x"${PID}" = x"" && {
echo "Usage: ${BASENAME} <pid> [<binary>]"
exit 1
}
DB_LIST="gdb"
case "${UNAME}" in
#
# on Tru64 we need to try ladebug first
# because gdb crashes itself...
#
OSF1)
DB_LIST="ladebug ${DB_LIST}"
;;
esac
for DB in ${DB_LIST}; do
DB_BIN=`which ${DB} 2>/dev/null | grep '^/'`
test x"${DB_BIN}" != x"" && {
break
}
done
test x"${DB_BIN}" = x"" && {
echo "${BASENAME}: ERROR: No debugger found."
exit 1
}
#
# we first try to use /proc/${PID}/exe
# then fallback to the binary from the commandline
# then we search for the commandline argument with
# 'which'
#
test -f "/proc/${PID}/exe" && BINARY="/proc/${PID}/exe"
test x"${BINARY}" = x"" && BINARY="/proc/${PID}/exe"
test -f "${BINARY}" || BINARY=`which ${BINARY}`
test -f "${BINARY}" || {
echo "${BASENAME}: ERROR: Cannot find binary '${BINARY}'."
exit 1
}
echo "${BASENAME}: Trying to use ${DB_BIN} on ${BINARY} on PID ${PID}"
BATCHFILE_PRE=/tmp/gdb_backtrace_pre.$$
BATCHFILE_MAIN=/tmp/gdb_backtrace_main.$$
case "${DB}" in
ladebug)
cat << EOF > ${BATCHFILE_PRE}
set \$stoponattach
EOF
cat << EOF > ${BATCHFILE_MAIN}
where
quit
EOF
${DB_BIN} -c "${BATCHFILE_MAIN}" -i "${BATCHFILE_PRE}" -pid "${PID}" "${BINARY}"
;;
gdb)
cat << EOF > ${BATCHFILE_MAIN}
set height 1000
bt full
quit
EOF
${DB_BIN} -x "${BATCHFILE_MAIN}" "${BINARY}" "${PID}"
;;
esac
/bin/rm -f ${BATCHFILE_PRE} ${BATCHFILE_MAIN}

73
ctdb/config/interface_modify.sh Executable file
View File

@ -0,0 +1,73 @@
#!/bin/sh
#
OP=$1
IFACE=$2
IP=$3
MASKBITS=$4
add_ip_to_iface()
{
local _iface=$1
local _ip=$2
local _maskbits=$3
# we make sure the interface is up first
/sbin/ip link set $_iface up || {
echo "Failed to bringup interface $_iface"
return 1;
}
/sbin/ip addr add $_ip/$_maskbits brd + dev $_iface || {
echo "Failed to add $_ip/$_maskbits on dev $_iface"
return 1;
}
return 0;
}
delete_ip_from_iface()
{
local _iface=$1
local _ip=$2
local _maskbits=$3
# the ip tool will delete all secondary IPs if this is the primary. To work around
# this _very_ annoying behaviour we have to keep a record of the secondaries and re-add
# them afterwards. yuck
local _secondaries=""
if /sbin/ip addr list dev $_iface primary | grep -q "inet $_ip/$_maskbits " ; then
_secondaries=`/sbin/ip addr list dev $_iface secondary | grep " inet " | awk '{print $2}'`
fi
local _failed=0
/sbin/ip addr del $_ip/$_maskbits dev $_iface || _failed=1
[ -z "$_secondaries" ] || {
local _i=""
for _i in $_secondaries; do
if /sbin/ip addr list dev $_iface | grep -q "inet $_i" ; then
echo "kept secondary $_i on dev $_iface"
else
echo "re-adding secondary address $_i to dev $_iface"
/sbin/ip addr add $_i dev $_iface || _failed=1
fi
done
}
[ $_failed = 0 ] || {
echo "Failed to del $_ip on dev $_iface"
return 1;
}
return 0;
}
case "$OP" in
add)
add_ip_to_iface $IFACE $IP $MASKBITS
exit $?
;;
delete)
delete_ip_from_iface $IFACE $IP $MASKBITS
exit $?
;;
esac
echo "$0: unknown operation[$OP]"
exit 1

7
ctdb/config/notify.sh
View File

@ -33,7 +33,12 @@ case $event in
# or do something else ...
;;
startup)
# do some extra magic when ctdb has started?
# do some extra magic when ctdb has finished the initial
# recovery?
;;
init)
# do some extra magic when ctdb has started?
;;
esac

10
ctdb/configure.ac
View File

@ -1,13 +1,5 @@
AC_PREREQ(2.50)
AC_INIT(ctdb, m4_esyscmd([grep 'Version:' ./packaging/RPM/ctdb.spec 2>/dev/null | head -1 | sed -e 's/[ \t]*Version:[ \t]*\([^ \t]*\)[ \t]*.*/\1/' | tr -d '\n']))
AC_DEFUN([AC_CHECK_LIB_EXT], [
AC_CHECK_LIB([$1],[$3],[$4],[$5],[$7])
ac_cv_lib_ext_$1_$3=$ac_cv_lib_$1_$3
])
AC_DEFUN([AC_CHECK_FUNC_EXT], [
AC_CHECK_FUNC([$1],[$3],[$4])
ac_cv_func_ext_$1=$ac_cv_func_$1
])
AC_DEFUN([SMB_MODULE_DEFAULT], [echo -n ""])
AC_DEFUN([SMB_LIBRARY_ENABLE], [echo -n ""])
AC_DEFUN([SMB_EXT_LIB], [echo -n ""])
@ -68,6 +60,8 @@ m4_include(libtalloc.m4)
m4_include(libtdb.m4)
m4_include(libevents.m4)
m4_include(ib/config.m4)
m4_include(lib/util/signal.m4)
m4_include(lib/util/fault.m4)
AC_CHECK_HEADERS(sched.h)
AC_CHECK_FUNCS(sched_setscheduler)

4
ctdb/include/ctdb.h
View File

@ -679,7 +679,8 @@ struct ctdb_scripts_wire {
/* different calls to event scripts. */
enum ctdb_eventscript_call {
CTDB_EVENT_STARTUP, /* CTDB starting up: no args. */
CTDB_EVENT_INIT, /* CTDB starting up: no args */
CTDB_EVENT_STARTUP, /* CTDB starting up after initial recovery: no args. */
CTDB_EVENT_START_RECOVERY, /* CTDB recovery starting: no args. */
CTDB_EVENT_RECOVERED, /* CTDB recovery finished: no args. */
CTDB_EVENT_TAKE_IP, /* IP taken: interface, IP address, netmask bits. */
@ -689,6 +690,7 @@ enum ctdb_eventscript_call {
CTDB_EVENT_STATUS, /* Report service status: no args. */
CTDB_EVENT_SHUTDOWN, /* CTDB shutting down: no args. */
CTDB_EVENT_RELOAD, /* magic */
CTDB_EVENT_UPDATE_IP, /* IP updating: old interface, new interface, IP address, netmask bits. */
CTDB_EVENT_MAX
};

77
ctdb/include/ctdb_private.h
View File

@ -181,12 +181,14 @@ struct ctdb_client {
struct ctdb_client_notify_list *notify;
};
struct ctdb_iface;
/* state associated with a public ip address */
struct ctdb_vnn {
struct ctdb_vnn *prev, *next;
const char *iface;
struct ctdb_iface *iface;
const char **ifaces;
ctdb_sock_addr public_address;
uint8_t public_netmask_bits;
@ -244,7 +246,8 @@ struct ctdb_node {
across the nodes. it needs to know which public ip's can be handled
by each node.
*/
struct ctdb_all_public_ips *public_ips;
struct ctdb_all_public_ips *known_public_ips;
struct ctdb_all_public_ips *available_public_ips;
/* used by the recovery dameon to track when a node should be banned */
struct ctdb_banning_state *ban_state;
};
@ -424,6 +427,7 @@ struct ctdb_context {
struct ctdb_node **nodes; /* array of nodes in the cluster - indexed by vnn */
struct ctdb_vnn *vnn; /* list of public ip addresses and interfaces */
struct ctdb_vnn *single_ip_vnn; /* a structure for the single ip */
struct ctdb_iface *ifaces; /* list of local interfaces */
char *err_msg;
const struct ctdb_methods *methods; /* transport methods */
const struct ctdb_upcalls *upcalls; /* transport upcalls */
@ -634,6 +638,9 @@ enum ctdb_controls {CTDB_CONTROL_PROCESS_EXISTS = 0,
CTDB_CONTROL_GET_DB_SEQNUM = 120,
CTDB_CONTROL_DB_SET_HEALTHY = 121,
CTDB_CONTROL_DB_GET_HEALTH = 122,
CTDB_CONTROL_GET_PUBLIC_IP_INFO = 123,
CTDB_CONTROL_GET_IFACES = 124,
CTDB_CONTROL_SET_IFACE_LINK_STATE = 125,
};
/*
@ -841,6 +848,7 @@ struct ctdb_req_control {
uint64_t srvid;
uint32_t client_id;
#define CTDB_CTRL_FLAG_NOREPLY 1
#define CTDB_CTRL_FLAG_OPCODE_SPECIFIC 0xFFFF0000
uint32_t flags;
uint32_t datalen;
uint8_t data[1];
@ -1301,12 +1309,67 @@ struct ctdb_all_public_ips {
int32_t ctdb_control_get_public_ipsv4(struct ctdb_context *ctdb, struct ctdb_req_control *c, TDB_DATA *outdata);
int32_t ctdb_control_get_public_ips(struct ctdb_context *ctdb, struct ctdb_req_control *c, TDB_DATA *outdata);
int ctdb_ctrl_get_public_ips(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx, struct ctdb_all_public_ips **ips);
struct timeval timeout,
uint32_t destnode,
TALLOC_CTX *mem_ctx,
struct ctdb_all_public_ips **ips);
#define CTDB_PUBLIC_IP_FLAGS_ONLY_AVAILABLE 0x00010000
int ctdb_ctrl_get_public_ips_flags(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx,
uint32_t flags,
struct ctdb_all_public_ips **ips);
int ctdb_ctrl_get_public_ipsv4(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx, struct ctdb_all_public_ips **ips);
#ifdef IFNAMSIZ
#define CTDB_IFACE_SIZE IFNAMSIZ
#else
#define CTDB_IFACE_SIZE 16
#endif
struct ctdb_control_iface_info {
char name[CTDB_IFACE_SIZE+2];
uint16_t link_state;
uint32_t references;
};
struct ctdb_control_public_ip_info {
struct ctdb_public_ip ip;
uint32_t active_idx;
uint32_t num;
struct ctdb_control_iface_info ifaces[1];
};
struct ctdb_control_get_ifaces {
uint32_t num;
struct ctdb_control_iface_info ifaces[1];
};
int32_t ctdb_control_get_public_ip_info(struct ctdb_context *ctdb,
struct ctdb_req_control *c,
TDB_DATA indata,
TDB_DATA *outdata);
int32_t ctdb_control_get_ifaces(struct ctdb_context *ctdb,
struct ctdb_req_control *c,
TDB_DATA *outdata);
int32_t ctdb_control_set_iface_link(struct ctdb_context *ctdb,
struct ctdb_req_control *c,
TDB_DATA indata);
int ctdb_ctrl_get_public_ip_info(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx,
const ctdb_sock_addr *addr,
struct ctdb_control_public_ip_info **info);
int ctdb_ctrl_get_ifaces(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx,
struct ctdb_control_get_ifaces **ifaces);
int ctdb_ctrl_set_iface_link(struct ctdb_context *ctdb,
struct timeval timeout, uint32_t destnode,
TALLOC_CTX *mem_ctx,
const struct ctdb_control_iface_info *info);
/* from takeover/system.c */
uint32_t uint16_checksum(uint16_t *data, size_t n);
@ -1317,6 +1380,9 @@ int ctdb_sys_send_tcp(const ctdb_sock_addr *dest,
uint32_t seq, uint32_t ack, int rst);
int ctdb_set_public_addresses(struct ctdb_context *ctdb, const char *alist);
int ctdb_set_single_public_ip(struct ctdb_context *ctdb,
const char *iface,
const char *ip);
int ctdb_set_event_script(struct ctdb_context *ctdb, const char *script);
int ctdb_set_event_script_dir(struct ctdb_context *ctdb, const char *script_dir);
int ctdb_set_notification_script(struct ctdb_context *ctdb, const char *script);
@ -1564,4 +1630,7 @@ int ctdb_update_persistent_health(struct ctdb_context *ctdb,
int num_healthy_nodes);
int ctdb_recheck_persistent_health(struct ctdb_context *ctdb);
void ctdb_run_notification_script(struct ctdb_context *ctdb, const char *event);
void ctdb_fault_setup(void);
#endif

7
ctdb/include/includes.h
View File

@ -7,7 +7,6 @@
#include "tdb.h"
#include "idtree.h"
#include "ctdb.h"
#include "lib/util/debug.h"
typedef bool BOOL;
@ -28,9 +27,13 @@ enum debug_level {
DEBUG_DEBUG = 4,
};
#define DEBUGLVL(lvl) ((lvl) <= LogLevel)
#define DEBUG(lvl, x) do { this_log_level = (lvl); if ((lvl) < DEBUG_DEBUG) { log_ringbuffer x; } if ((lvl) <= LogLevel) { do_debug x; }} while (0)
#define DEBUGADD(lvl, x) do { if ((lvl) <= LogLevel) { this_log_level = (lvl); do_debug_add x; }} while (0)
#define _PUBLIC_
#define _NORETURN_
#define _PURE_
#define ZERO_STRUCT(x) memset((char *)&(x), 0, sizeof(x))
@ -53,3 +56,5 @@ char *hex_encode_talloc(TALLOC_CTX *mem_ctx, const unsigned char *buff_in, size_
_PUBLIC_ const char **str_list_add(const char **list, const char *s);
_PUBLIC_ int set_blocking(int fd, BOOL set);
#include "lib/util/debug.h"
#include "lib/util/util.h"

10
ctdb/lib/events/events_epoll.c
View File

@ -41,14 +41,6 @@ struct epoll_event_context {
/* number of registered fd event handlers */
int num_fd_events;
/* this is changed by the destructors for the fd event
type. It is used to detect event destruction by event
handlers, which means the code that is calling the event
handler needs to assume that the linked list is no longer
valid
*/
uint32_t destruction_count;
/* when using epoll this is the handle from epoll_create */
int epoll_fd;
@ -260,7 +252,6 @@ static int epoll_event_loop(struct epoll_event_context *epoll_ev, struct timeval
int ret, i;
#define MAXEVENTS 32
struct epoll_event events[MAXEVENTS];
uint32_t destruction_count = ++epoll_ev->destruction_count;
int timeout = -1;
if (epoll_ev->epoll_fd == -1) return -1;
@ -358,7 +349,6 @@ static int epoll_event_fd_destructor(struct fd_event *fde)
epoll_check_reopen(epoll_ev);
epoll_ev->num_fd_events--;
epoll_ev->destruction_count++;
epoll_del_event(epoll_ev, fde);

6
ctdb/lib/events/events_select.c
View File

@ -48,10 +48,6 @@ struct select_event_context {
/* information for exiting from the event loop */
int exit_code;
/* this is incremented when the loop over events causes something which
could change the events yet to be processed */
uint32_t destruction_count;
};
/*
@ -104,7 +100,6 @@ static int select_event_fd_destructor(struct fd_event *fde)
}
DLIST_REMOVE(select_ev->fd_events, fde);
select_ev->destruction_count++;
if (fde->flags & EVENT_FD_AUTOCLOSE) {
close(fde->fd);
@ -180,7 +175,6 @@ static int select_event_loop_select(struct select_event_context *select_ev, stru
fd_set r_fds, w_fds;
struct fd_event *fde;
int selrtn;
uint32_t destruction_count = ++select_ev->destruction_count;
/* we maybe need to recalculate the maxfd */
if (select_ev->maxfd == EVENT_INVALID_MAXFD) {

19
ctdb/lib/events/events_standard.c
View File

@ -51,14 +51,6 @@ struct std_event_context {
/* information for exiting from the event loop */
int exit_code;
/* this is changed by the destructors for the fd event
type. It is used to detect event destruction by event
handlers, which means the code that is calling the event
handler needs to assume that the linked list is no longer
valid
*/
uint32_t destruction_count;
/* when using epoll this is the handle from epoll_create */
int epoll_fd;
@ -255,7 +247,6 @@ static int epoll_event_loop(struct std_event_context *std_ev, struct timeval *tv
int ret, i;
#define MAXEVENTS 8
struct epoll_event events[MAXEVENTS];
uint32_t destruction_count = ++std_ev->destruction_count;
int timeout = -1;
if (std_ev->epoll_fd == -1) return -1;
@ -316,9 +307,7 @@ static int epoll_event_loop(struct std_event_context *std_ev, struct timeval *tv
if (events[i].events & EPOLLOUT) flags |= EVENT_FD_WRITE;
if (flags) {
fde->handler(std_ev->ev, fde, flags, fde->private_data);
if (destruction_count != std_ev->destruction_count) {
break;
}
break;
}
}
@ -388,7 +377,6 @@ static int std_event_fd_destructor(struct fd_event *fde)
}
DLIST_REMOVE(std_ev->fd_events, fde);
std_ev->destruction_count++;
epoll_del_event(std_ev, fde);
@ -475,7 +463,6 @@ static int std_event_loop_select(struct std_event_context *std_ev, struct timeva
fd_set r_fds, w_fds;
struct fd_event *fde;
int selrtn;
uint32_t destruction_count = ++std_ev->destruction_count;
/* we maybe need to recalculate the maxfd */
if (std_ev->maxfd == EVENT_INVALID_MAXFD) {
@ -536,9 +523,7 @@ static int std_event_loop_select(struct std_event_context *std_ev, struct timeva
if (FD_ISSET(fde->fd, &w_fds)) flags |= EVENT_FD_WRITE;
if (flags) {
fde->handler(std_ev->ev, fde, flags, fde->private_data);
if (destruction_count != std_ev->destruction_count) {
break;
}
break;
}
}
}

4
ctdb/lib/util/debug.c
View File

@ -89,10 +89,6 @@ void do_debug_add(const char *format, ...)
va_end(ap);
}
#define DEBUGLVL(lvl) ((lvl) <= LogLevel)
#define DEBUG(lvl, x) do { this_log_level = (lvl); if ((lvl) < DEBUG_DEBUG) { log_ringbuffer x; } if ((lvl) <= LogLevel) { do_debug x; }} while (0)
#define DEBUGADD(lvl, x) do { if ((lvl) <= LogLevel) { this_log_level = (lvl); do_debug_add x; }} while (0)
static void print_asc(int level, const uint8_t *buf, size_t len)
{
int i;

2
ctdb/lib/util/debug.h
View File

@ -21,4 +21,6 @@ void (*do_debug_v)(const char *, va_list ap);
void (*do_debug_add_v)(const char *, va_list ap);
void log_ringbuffer(const char *format, ...);
void do_debug(const char *format, ...) PRINTF_ATTRIBUTE(1, 2);
void do_debug_add(const char *format, ...) PRINTF_ATTRIBUTE(1, 2);
void dump_data(int level, const uint8_t *buf1, size_t len);

239
ctdb/lib/util/fault.c Normal file
View File

@ -0,0 +1,239 @@
/*
Unix SMB/CIFS implementation.
Critical Fault handling
Copyright (C) Andrew Tridgell 1992-1998
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 3 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, see <http://www.gnu.org/licenses/>.
*/
#include "includes.h"
#include "system/wait.h"
#include "system/filesys.h"
/**
* @file
* @brief Fault handling
*/
/* the registered fault handler */
static struct {
const char *name;
void (*fault_handler)(int sig);
} fault_handlers;
static const char *progname;
#ifdef HAVE_BACKTRACE
#include <execinfo.h>
#elif HAVE_LIBEXC_H
#include <libexc.h>
#endif
/**
* Write backtrace to debug log
*/
_PUBLIC_ void call_backtrace(void)
{
#ifdef HAVE_BACKTRACE
#ifndef BACKTRACE_STACK_SIZE
#define BACKTRACE_STACK_SIZE 64
#endif
void *backtrace_stack[BACKTRACE_STACK_SIZE];
size_t backtrace_size;
char **backtrace_strings;
/* get the backtrace (stack frames) */
backtrace_size = backtrace(backtrace_stack,BACKTRACE_STACK_SIZE);
backtrace_strings = backtrace_symbols(backtrace_stack, backtrace_size);
DEBUG(0, ("BACKTRACE: %lu stack frames:\n",
(unsigned long)backtrace_size));
if (backtrace_strings) {
int i;
for (i = 0; i < backtrace_size; i++)
DEBUGADD(0, (" #%u %s\n", i, backtrace_strings[i]));
/* Leak the backtrace_strings, rather than risk what free() might do */
}
#elif HAVE_LIBEXC
#define NAMESIZE 32 /* Arbitrary */
#ifndef BACKTRACE_STACK_SIZE
#define BACKTRACE_STACK_SIZE 64
#endif
/* The IRIX libexc library provides an API for unwinding the stack. See
* libexc(3) for details. Apparantly trace_back_stack leaks memory, but
* since we are about to abort anyway, it hardly matters.
*
* Note that if we paniced due to a SIGSEGV or SIGBUS (or similar) this
* will fail with a nasty message upon failing to open the /proc entry.
*/
{
uint64_t addrs[BACKTRACE_STACK_SIZE];
char * names[BACKTRACE_STACK_SIZE];
char namebuf[BACKTRACE_STACK_SIZE * NAMESIZE];
int i;
int levels;
ZERO_ARRAY(addrs);
ZERO_ARRAY(names);
ZERO_ARRAY(namebuf);
for (i = 0; i < BACKTRACE_STACK_SIZE; i++) {
names[i] = namebuf + (i * NAMESIZE);
}
levels = trace_back_stack(0, addrs, names,
BACKTRACE_STACK_SIZE, NAMESIZE);
DEBUG(0, ("BACKTRACE: %d stack frames:\n", levels));
for (i = 0; i < levels; i++) {
DEBUGADD(0, (" #%d 0x%llx %s\n", i, addrs[i], names[i]));
}
}
#undef NAMESIZE
#else
DEBUG(0, ("call_backstrace: not implemented\n"));
#error bla
#endif
}
_PUBLIC_ const char *panic_action = NULL;
_PUBLIC_ void (*pre_panic_action_hook)(void) = NULL;
_PUBLIC_ void (*post_panic_action_hook)(void) = NULL;
/**
Something really nasty happened - panic !
**/
_PUBLIC_ void smb_panic(const char *why)
{
int result;
if (panic_action && *panic_action) {
char pidstr[20];
char cmdstring[200];
strlcpy(cmdstring, panic_action, sizeof(cmdstring));
snprintf(pidstr, sizeof(pidstr), "%u", getpid());
all_string_sub(cmdstring, "%PID%", pidstr, sizeof(cmdstring));
if (progname) {
all_string_sub(cmdstring, "%PROG%", progname, sizeof(cmdstring));
}
DEBUG(0, ("smb_panic(): calling panic action [%s]\n", cmdstring));
if (pre_panic_action_hook) {
pre_panic_action_hook();
}
result = system(cmdstring);
if (post_panic_action_hook) {
post_panic_action_hook();
}
if (result == -1)
DEBUG(0, ("smb_panic(): fork failed in panic action: %s\n",
strerror(errno)));
else
DEBUG(0, ("smb_panic(): action returned status %d\n",
WEXITSTATUS(result)));
}
DEBUG(0,("PANIC: %s\n", why));
call_backtrace();
#ifdef SIGABRT
CatchSignal(SIGABRT,SIGNAL_CAST SIG_DFL);
#endif
abort();
}
/**
report a fault
**/
_NORETURN_ static void fault_report(int sig)
{
static int counter;
if (counter) _exit(1);
DEBUG(0,("=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=\n"));
DEBUG(0,("INTERNAL ERROR: Signal %d in %s pid %d",sig, progname, (int)getpid()));
DEBUG(0,("\nPlease read the file BUGS.txt in the distribution\n"));
DEBUG(0,("=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=\n"));
smb_panic("internal error");
exit(1);
}
/**
catch serious errors
**/
_NORETURN_ static void sig_fault(int sig)
{
if (fault_handlers.fault_handler) {
/* we have a fault handler, call it. It may not return. */
fault_handlers.fault_handler(sig);
}
/* If it returns or doesn't exist, use regular reporter */
fault_report(sig);
}
/**
setup our fault handlers
**/
_PUBLIC_ void fault_setup(const char *pname)
{
if (progname == NULL) {
progname = pname;
}
#ifdef SIGSEGV
CatchSignal(SIGSEGV,SIGNAL_CAST sig_fault);
#endif
#ifdef SIGBUS
CatchSignal(SIGBUS,SIGNAL_CAST sig_fault);
#endif
#ifdef SIGABRT
CatchSignal(SIGABRT,SIGNAL_CAST sig_fault);
#endif
#ifdef SIGFPE
CatchSignal(SIGFPE,SIGNAL_CAST sig_fault);
#endif
}
/**
register a fault handler.
Should only be called once in the execution of smbd.
*/
_PUBLIC_ bool register_fault_handler(const char *name,
void (*fault_handler)(int sig))
{
if (fault_handlers.name != NULL) {
/* it's already registered! */
DEBUG(2,("fault handler '%s' already registered - failed '%s'\n",
fault_handlers.name, name));
return false;
}
fault_handlers.name = name;
fault_handlers.fault_handler = fault_handler;
DEBUG(2,("fault handler '%s' registered\n", name));
return true;
}

15
ctdb/lib/util/fault.m4 Normal file
View File

@ -0,0 +1,15 @@
AC_CHECK_HEADERS(execinfo.h)
AC_SEARCH_LIBS_EXT(backtrace, [execinfo], EXECINFO_LIBS)
AC_CHECK_FUNC_EXT(backtrace, $EXECINFO_LIBS)
if test x"$ac_cv_header_execinfo_h" = x"yes" -a x"$ac_cv_func_ext_backtrace" = x"yes";then
SMB_ENABLE(EXECINFO, YES)
EXECINFO_CFLAGS="$CFLAGS"
EXECINFO_CPPFLAGS="$CPPFLAGS"
EXECINFO_LDFLAGS="$LDFLAGS"
else
SMB_ENABLE(EXECINFO,NO)
fi
SMB_EXT_LIB(EXECINFO, [${EXECINFO_LIBS}], [${EXECINFO_CFLAGS}], [${EXECINFO_CPPFLAGS}], [${EXECINFO_LDFLAGS}])

144
ctdb/lib/util/signal.c Normal file
View File

@ -0,0 +1,144 @@
/*
Unix SMB/CIFS implementation.
signal handling functions
Copyright (C) Andrew Tridgell 1998
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 3 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, see <http://www.gnu.org/licenses/>.
*/
#include "includes.h"
#include "system/wait.h"
/**
* @file
* @brief Signal handling
*/
/****************************************************************************
Catch child exits and reap the child zombie status.
****************************************************************************/
static void sig_cld(int signum)
{
while (waitpid((pid_t)-1,(int *)NULL, WNOHANG) > 0)
;
/*
* Turns out it's *really* important not to
* restore the signal handler here if we have real POSIX
* signal handling. If we do, then we get the signal re-delivered
* immediately - hey presto - instant loop ! JRA.
*/
#if !defined(HAVE_SIGACTION)
CatchSignal(SIGCLD, sig_cld);
#endif
}
/****************************************************************************
catch child exits - leave status;
****************************************************************************/
static void sig_cld_leave_status(int signum)
{
/*
* Turns out it's *really* important not to
* restore the signal handler here if we have real POSIX
* signal handling. If we do, then we get the signal re-delivered
* immediately - hey presto - instant loop ! JRA.
*/
#if !defined(HAVE_SIGACTION)
CatchSignal(SIGCLD, sig_cld_leave_status);
#endif
}
/**
Block sigs.
**/
void BlockSignals(bool block, int signum)
{
#ifdef HAVE_SIGPROCMASK
sigset_t set;
sigemptyset(&set);
sigaddset(&set,signum);
sigprocmask(block?SIG_BLOCK:SIG_UNBLOCK,&set,NULL);
#elif defined(HAVE_SIGBLOCK)
if (block) {
sigblock(sigmask(signum));
} else {
sigsetmask(siggetmask() & ~sigmask(signum));
}
#else
/* yikes! This platform can't block signals? */
static int done;
if (!done) {
DEBUG(0,("WARNING: No signal blocking available\n"));
done=1;
}
#endif
}
/**
Catch a signal. This should implement the following semantics:
1) The handler remains installed after being called.
2) The signal should be blocked during handler execution.
**/
void (*CatchSignal(int signum,void (*handler)(int )))(int)
{
#ifdef HAVE_SIGACTION
struct sigaction act;
struct sigaction oldact;
ZERO_STRUCT(act);
act.sa_handler = handler;
#ifdef SA_RESTART
/*
* We *want* SIGALRM to interrupt a system call.
*/
if(signum != SIGALRM)
act.sa_flags = SA_RESTART;
#endif
sigemptyset(&act.sa_mask);
sigaddset(&act.sa_mask,signum);
sigaction(signum,&act,&oldact);
return oldact.sa_handler;
#else /* !HAVE_SIGACTION */
/* FIXME: need to handle sigvec and systems with broken signal() */
return signal(signum, handler);
#endif
}
/**
Ignore SIGCLD via whatever means is necessary for this OS.
**/
void CatchChild(void)
{
CatchSignal(SIGCLD, sig_cld);
}
/**
Catch SIGCLD but leave the child around so it's status can be reaped.
**/
void CatchChildLeaveStatus(void)
{
CatchSignal(SIGCLD, sig_cld_leave_status);
}

1
ctdb/lib/util/signal.m4 Normal file
View File

@ -0,0 +1 @@
AC_CHECK_FUNCS(sigprocmask sigblock sigaction)

167
ctdb/lib/util/substitute.c Normal file
View File

@ -0,0 +1,167 @@
/*
Unix SMB/CIFS implementation.
Samba utility functions
Copyright (C) Andrew Tridgell 1992-2001
Copyright (C) Simo Sorce 2001-2002
Copyright (C) Martin Pool 2003
Copyright (C) James Peach 2005
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 3 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, see <http://www.gnu.org/licenses/>.
*/
#include "includes.h"
/**
* @file
* @brief Substitute utilities.
**/
/**
Substitute a string for a pattern in another string. Make sure there is
enough room!
This routine looks for pattern in s and replaces it with
insert. It may do multiple replacements.
Any of " ; ' $ or ` in the insert string are replaced with _
if len==0 then the string cannot be extended. This is different from the old
use of len==0 which was for no length checks to be done.
**/
_PUBLIC_ void string_sub(char *s, const char *pattern, const char *insert, size_t len)
{
char *p;
ssize_t ls, lp, li, i;
if (!insert || !pattern || !*pattern || !s)
return;
ls = (ssize_t)strlen(s);
lp = (ssize_t)strlen(pattern);
li = (ssize_t)strlen(insert);
if (len == 0)
len = ls + 1; /* len is number of *bytes* */
while (lp <= ls && (p = strstr(s, pattern))) {
if (ls + (li-lp) >= len) {
DEBUG(0,("ERROR: string overflow by %d in string_sub(%.50s, %d)\n",
(int)(ls + (li-lp) - len),
pattern, (int)len));
break;
}
if (li != lp) {
memmove(p+li,p+lp,strlen(p+lp)+1);
}
for (i=0;i<li;i++) {
switch (insert[i]) {
case '`':
case '"':
case '\'':
case ';':
case '$':
case '%':
case '\r':
case '\n':
p[i] = '_';
break;
default:
p[i] = insert[i];
}
}
s = p + li;
ls += (li-lp);
}
}
/**
* Talloc'ed version of string_sub
*/
_PUBLIC_ char *string_sub_talloc(TALLOC_CTX *mem_ctx, const char *s,
const char *pattern, const char *insert)
{
const char *p;
char *ret;
size_t len, alloc_len;
if (insert == NULL || pattern == NULL || !*pattern || s == NULL)
return NULL;
/* determine length needed */
len = strlen(s);
for (p = strstr(s, pattern); p != NULL;
p = strstr(p+strlen(pattern), pattern)) {
len += strlen(insert) - strlen(pattern);
}
alloc_len = MAX(len, strlen(s))+1;
ret = talloc_array(mem_ctx, char, alloc_len);
if (ret == NULL)
return NULL;
strncpy(ret, s, alloc_len);
string_sub(ret, pattern, insert, alloc_len);
ret = talloc_realloc(mem_ctx, ret, char, len+1);
if (ret == NULL)
return NULL;
SMB_ASSERT(ret[len] == '\0');
talloc_set_name_const(ret, ret);
return ret;
}
/**
Similar to string_sub() but allows for any character to be substituted.
Use with caution!
if len==0 then the string cannot be extended. This is different from the old
use of len==0 which was for no length checks to be done.
**/
_PUBLIC_ void all_string_sub(char *s,const char *pattern,const char *insert, size_t len)
{
char *p;
ssize_t ls,lp,li;
if (!insert || !pattern || !s)
return;
ls = (ssize_t)strlen(s);
lp = (ssize_t)strlen(pattern);
li = (ssize_t)strlen(insert);
if (!*pattern)
return;
if (len == 0)
len = ls + 1; /* len is number of *bytes* */
while (lp <= ls && (p = strstr(s,pattern))) {
if (ls + (li-lp) >= len) {
DEBUG(0,("ERROR: string overflow by %d in all_string_sub(%.50s, %d)\n",
(int)(ls + (li-lp) - len),
pattern, (int)len));
break;
}
if (li != lp) {
memmove(p+li,p+lp,strlen(p+lp)+1);
}
memcpy(p, insert, li);
s = p + li;
ls += (li-lp);
}
}

658
ctdb/lib/util/util.h Normal file
View File

@ -0,0 +1,658 @@
/*
Unix SMB/CIFS implementation.
Utility functions for Samba
Copyright (C) Andrew Tridgell 1992-1999
Copyright (C) Jelmer Vernooij 2005
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 3 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, see <http://www.gnu.org/licenses/>.
*/
#ifndef _SAMBA_UTIL_H_
#define _SAMBA_UTIL_H_
/**
* @file
* @brief Helpful macros
*/
struct smbsrv_tcon;
#ifdef _SAMBA_BUILD_
extern const char *logfile;
#endif
extern const char *panic_action;
extern void (*pre_panic_action_hook)(void);
extern void (*post_panic_action_hook)(void);
/**
* assert macros
*/
#ifdef DEVELOPER
#define SMB_ASSERT(b) do { if (!(b)) { \
DEBUG(0,("PANIC: assert failed at %s(%d): %s\n", \
__FILE__, __LINE__, #b)); smb_panic("assert failed: " #b); }} while(0)
#else
/* redefine the assert macro for non-developer builds */
#define SMB_ASSERT(b) do { if (!(b)) { \
DEBUG(0,("PANIC: assert failed at %s(%d): %s\n", \
__FILE__, __LINE__, #b)); }} while (0)
#endif
#if _SAMBA_BUILD_ == 4
#ifdef VALGRIND
#define strlen(x) valgrind_strlen(x)
size_t valgrind_strlen(const char *s);
#endif
#endif
#ifndef ABS
#define ABS(a) ((a)>0?(a):(-(a)))
#endif
/**
* Write backtrace to debug log
*/
_PUBLIC_ void call_backtrace(void);
/**
Something really nasty happened - panic !
**/
_PUBLIC_ _NORETURN_ void smb_panic(const char *why);
/**
setup our fault handlers
**/
_PUBLIC_ void fault_setup(const char *pname);
/**
register a fault handler.
Should only be called once in the execution of smbd.
*/
_PUBLIC_ bool register_fault_handler(const char *name, void (*fault_handler)(int sig));
/* The following definitions come from lib/util/signal.c */
/**
Block sigs.
**/
void BlockSignals(bool block, int signum);
/**
Catch a signal. This should implement the following semantics:
1) The handler remains installed after being called.
2) The signal should be blocked during handler execution.
**/
void (*CatchSignal(int signum,void (*handler)(int )))(int);
/**
Ignore SIGCLD via whatever means is necessary for this OS.
**/
void CatchChild(void);
/**
Catch SIGCLD but leave the child around so it's status can be reaped.
**/
void CatchChildLeaveStatus(void);
/* The following definitions come from lib/util/util_str.c */
/**
Trim the specified elements off the front and back of a string.
**/
_PUBLIC_ bool trim_string(char *s, const char *front, const char *back);
/**
Find the number of 'c' chars in a string
**/
_PUBLIC_ _PURE_ size_t count_chars(const char *s, char c);
/**
Safe string copy into a known length string. maxlength does not
include the terminating zero.
**/
_PUBLIC_ char *safe_strcpy(char *dest,const char *src, size_t maxlength);
/**
Safe string cat into a string. maxlength does not
include the terminating zero.
**/
_PUBLIC_ char *safe_strcat(char *dest, const char *src, size_t maxlength);
/**
Routine to get hex characters and turn them into a 16 byte array.
the array can be variable length, and any non-hex-numeric
characters are skipped. "0xnn" or "0Xnn" is specially catered
for.
valid examples: "0A5D15"; "0x15, 0x49, 0xa2"; "59\ta9\te3\n"
**/
_PUBLIC_ size_t strhex_to_str(char *p, size_t p_len, const char *strhex, size_t strhex_len);
#ifdef _SAMBA_BUILD_
/**
* Parse a hex string and return a data blob.
*/
_PUBLIC_ _PURE_ DATA_BLOB strhex_to_data_blob(TALLOC_CTX *mem_ctx, const char *strhex) ;
#endif
/**
* Routine to print a buffer as HEX digits, into an allocated string.
*/
_PUBLIC_ void hex_encode(const unsigned char *buff_in, size_t len, char **out_hex_buffer);
/**
* talloc version of hex_encode()
*/
_PUBLIC_ char *hex_encode_talloc(TALLOC_CTX *mem_ctx, const unsigned char *buff_in, size_t len);
/**
Substitute a string for a pattern in another string. Make sure there is
enough room!
This routine looks for pattern in s and replaces it with
insert. It may do multiple replacements.
Any of " ; ' $ or ` in the insert string are replaced with _
if len==0 then the string cannot be extended. This is different from the old
use of len==0 which was for no length checks to be done.
**/
_PUBLIC_ void string_sub(char *s,const char *pattern, const char *insert, size_t len);
_PUBLIC_ char *string_sub_talloc(TALLOC_CTX *mem_ctx, const char *s,
const char *pattern, const char *insert);
/**
Similar to string_sub() but allows for any character to be substituted.
Use with caution!
if len==0 then the string cannot be extended. This is different from the old
use of len==0 which was for no length checks to be done.
**/
_PUBLIC_ void all_string_sub(char *s,const char *pattern,const char *insert, size_t len);
/**
Unescape a URL encoded string, in place.
**/
_PUBLIC_ void rfc1738_unescape(char *buf);
/**
format a string into length-prefixed dotted domain format, as used in NBT
and in some ADS structures
**/
_PUBLIC_ const char *str_format_nbt_domain(TALLOC_CTX *mem_ctx, const char *s);
/**
* Add a string to an array of strings.
*
* num should be a pointer to an integer that holds the current
* number of elements in strings. It will be updated by this function.
*/
_PUBLIC_ bool add_string_to_array(TALLOC_CTX *mem_ctx,
const char *str, const char ***strings, int *num);
/**
varient of strcmp() that handles NULL ptrs
**/
_PUBLIC_ int strcmp_safe(const char *s1, const char *s2);
/**
return the number of bytes occupied by a buffer in ASCII format
the result includes the null termination
limited by 'n' bytes
**/
_PUBLIC_ size_t ascii_len_n(const char *src, size_t n);
/**
Set a boolean variable from the text value stored in the passed string.
Returns true in success, false if the passed string does not correctly
represent a boolean.
**/
_PUBLIC_ bool set_boolean(const char *boolean_string, bool *boolean);
/**
* Parse a string containing a boolean value.
*
* val will be set to the read value.
*
* @retval true if a boolean value was parsed, false otherwise.
*/
_PUBLIC_ bool conv_str_bool(const char * str, bool * val);
#if _SAMBA_BUILD_ == 4
/**
* Convert a size specification like 16K into an integral number of bytes.
**/
_PUBLIC_ bool conv_str_size(const char * str, uint64_t * val);
#endif
/**
* Parse a uint64_t value from a string
*
* val will be set to the value read.
*
* @retval true if parsing was successful, false otherwise
*/
_PUBLIC_ bool conv_str_u64(const char * str, uint64_t * val);
/**
return the number of bytes occupied by a buffer in CH_UTF16 format
the result includes the null termination
**/
_PUBLIC_ size_t utf16_len(const void *buf);
/**
return the number of bytes occupied by a buffer in CH_UTF16 format
the result includes the null termination
limited by 'n' bytes
**/
_PUBLIC_ size_t utf16_len_n(const void *src, size_t n);
_PUBLIC_ size_t ucs2_align(const void *base_ptr, const void *p, int flags);
/**
Do a case-insensitive, whitespace-ignoring string compare.
**/
_PUBLIC_ int strwicmp(const char *psz1, const char *psz2);
/**
String replace.
**/
_PUBLIC_ void string_replace(char *s, char oldc, char newc);
/**
* Compare 2 strings.
*
* @note The comparison is case-insensitive.
**/
_PUBLIC_ bool strequal(const char *s1, const char *s2);
/* The following definitions come from lib/util/util_strlist.c */
#ifdef _SAMBA_BUILD_
/* separators for lists */
#ifndef LIST_SEP
#define LIST_SEP " \t,\n\r"
#endif
/**
build a null terminated list of strings from a input string and a
separator list. The separator list must contain characters less than
or equal to 0x2f for this to work correctly on multi-byte strings
*/
_PUBLIC_ char **str_list_make(TALLOC_CTX *mem_ctx, const char *string, const char *sep);
/**
* build a null terminated list of strings from an argv-like input string
* Entries are seperated by spaces and can be enclosed by quotes.
* Does NOT support escaping
*/
_PUBLIC_ const char **str_list_make_shell(TALLOC_CTX *mem_ctx, const char *string, const char *sep);
/**
* join a list back to one string
*/
_PUBLIC_ char *str_list_join(TALLOC_CTX *mem_ctx, const char **list, char seperator);
/** join a list back to one (shell-like) string; entries
* seperated by spaces, using quotes where necessary */
_PUBLIC_ char *str_list_join_shell(TALLOC_CTX *mem_ctx, const char **list, char sep);
/**
return the number of elements in a string list
*/
_PUBLIC_ size_t str_list_length(const char * const *list);
/**
copy a string list
*/
_PUBLIC_ char **str_list_copy(TALLOC_CTX *mem_ctx, const char **list);
/**
Return true if all the elements of the list match exactly.
*/
_PUBLIC_ bool str_list_equal(const char **list1, const char **list2);
/**
add an entry to a string list
*/
_PUBLIC_ const char **str_list_add(const char **list, const char *s);
/**
remove an entry from a string list
*/
_PUBLIC_ void str_list_remove(const char **list, const char *s);
/**
return true if a string is in a list
*/
_PUBLIC_ bool str_list_check(const char **list, const char *s);
/**
return true if a string is in a list, case insensitively
*/
_PUBLIC_ bool str_list_check_ci(const char **list, const char *s);
#endif
/* The following definitions come from lib/util/util_file.c */
#ifdef _SAMBA_BUILD_
/**
read a line from a file with possible \ continuation chars.
Blanks at the start or end of a line are stripped.
The string will be allocated if s2 is NULL
**/
_PUBLIC_ char *fgets_slash(char *s2,int maxlen,XFILE *f);
#endif
/**
* Read one line (data until next newline or eof) and allocate it
*/
_PUBLIC_ char *afdgets(int fd, TALLOC_CTX *mem_ctx, size_t hint);
#ifdef _SAMBA_BUILD_
/**
load a file into memory from a fd.
**/
_PUBLIC_ char *fd_load(int fd, size_t *size, size_t maxsize, TALLOC_CTX *mem_ctx);
char **file_lines_parse(char *p, size_t size, int *numlines, TALLOC_CTX *mem_ctx);
/**
load a file into memory
**/
_PUBLIC_ char *file_load(const char *fname, size_t *size, size_t maxsize, TALLOC_CTX *mem_ctx);
#endif
/**
mmap (if possible) or read a file
**/
_PUBLIC_ void *map_file(const char *fname, size_t size);
#ifdef _SAMBA_BUILD_
/**
load a file into memory and return an array of pointers to lines in the file
must be freed with talloc_free().
**/
_PUBLIC_ char **file_lines_load(const char *fname, int *numlines, size_t maxsize, TALLOC_CTX *mem_ctx);
#endif
/**
load a fd into memory and return an array of pointers to lines in the file
must be freed with talloc_free(). If convert is true calls unix_to_dos on
the list.
**/
_PUBLIC_ char **fd_lines_load(int fd, int *numlines, size_t maxsize, TALLOC_CTX *mem_ctx);
/**
take a list of lines and modify them to produce a list where \ continues
a line
**/
_PUBLIC_ void file_lines_slashcont(char **lines);
/**
save a lump of data into a file. Mostly used for debugging
*/
_PUBLIC_ bool file_save(const char *fname, const void *packet, size_t length);
_PUBLIC_ int vfdprintf(int fd, const char *format, va_list ap) PRINTF_ATTRIBUTE(2,0);
_PUBLIC_ int fdprintf(int fd, const char *format, ...) PRINTF_ATTRIBUTE(2,3);
_PUBLIC_ bool large_file_support(const char *path);
/* The following definitions come from lib/util/util.c */
/**
Find a suitable temporary directory. The result should be copied immediately
as it may be overwritten by a subsequent call.
**/
_PUBLIC_ const char *tmpdir(void);
/**
Check if a file exists - call vfs_file_exist for samba files.
**/
_PUBLIC_ bool file_exist(const char *fname);
/**
Check a files mod time.
**/
_PUBLIC_ time_t file_modtime(const char *fname);
/**
Check if a directory exists.
**/
_PUBLIC_ bool directory_exist(const char *dname);
/**
* Try to create the specified directory if it didn't exist.
*
* @retval true if the directory already existed and has the right permissions
* or was successfully created.
*/
_PUBLIC_ bool directory_create_or_exist(const char *dname, uid_t uid,
mode_t dir_perms);
/**
Set a fd into blocking/nonblocking mode. Uses POSIX O_NONBLOCK if available,
else
if SYSV use O_NDELAY
if BSD use FNDELAY
**/
_PUBLIC_ int set_blocking(int fd, bool set);
/**
Sleep for a specified number of milliseconds.
**/
_PUBLIC_ void msleep(unsigned int t);
/**
Get my own name, return in malloc'ed storage.
**/
_PUBLIC_ char* get_myname(void);
/**
Return true if a string could be a pure IP address.
**/
_PUBLIC_ bool is_ipaddress(const char *str);
/**
Interpret an internet address or name into an IP address in 4 byte form.
**/
_PUBLIC_ uint32_t interpret_addr(const char *str);
/**
A convenient addition to interpret_addr().
**/
_PUBLIC_ struct in_addr interpret_addr2(const char *str);
/**
Check if an IP is the 0.0.0.0.
**/
_PUBLIC_ bool is_zero_ip_v4(struct in_addr ip);
/**
Are two IPs on the same subnet?
**/
_PUBLIC_ bool same_net_v4(struct in_addr ip1,struct in_addr ip2,struct in_addr mask);
_PUBLIC_ bool is_ipaddress_v4(const char *str);
/**
Check if a process exists. Does this work on all unixes?
**/
_PUBLIC_ bool process_exists_by_pid(pid_t pid);
/**
Simple routine to do POSIX file locking. Cruft in NFS and 64->32 bit mapping
is dealt with in posix.c
**/
_PUBLIC_ bool fcntl_lock(int fd, int op, off_t offset, off_t count, int type);
/**
malloc that aborts with smb_panic on fail or zero size.
**/
_PUBLIC_ void *smb_xmalloc(size_t size);
/**
Memdup with smb_panic on fail.
**/
_PUBLIC_ void *smb_xmemdup(const void *p, size_t size);
/**
strdup that aborts on malloc fail.
**/
_PUBLIC_ char *smb_xstrdup(const char *s);
char *smb_xstrndup(const char *s, size_t n);
/**
Like strdup but for memory.
**/
_PUBLIC_ void *memdup(const void *p, size_t size);
/**
* see if a range of memory is all zero. A NULL pointer is considered
* to be all zero
*/
_PUBLIC_ bool all_zero(const uint8_t *ptr, size_t size);
/**
realloc an array, checking for integer overflow in the array size
*/
_PUBLIC_ void *realloc_array(void *ptr, size_t el_size, unsigned count, bool free_on_fail);
void *malloc_array(size_t el_size, unsigned int count);
/* The following definitions come from lib/util/fsusage.c */
/**
* Retrieve amount of free disk space.
* this does all of the system specific guff to get the free disk space.
* It is derived from code in the GNU fileutils package, but has been
* considerably mangled for use here
*
* results are returned in *dfree and *dsize, in 512 byte units
*/
_PUBLIC_ int sys_fsusage(const char *path, uint64_t *dfree, uint64_t *dsize);
/* The following definitions come from lib/util/ms_fnmatch.c */
/**
* @file
* @brief MS-style Filename matching
*/
#if _SAMBA_BUILD_ == 4
/* protocol types. It assumes that higher protocols include lower protocols
as subsets. FIXME: Move to one of the smb-specific headers */
enum protocol_types {
PROTOCOL_NONE,
PROTOCOL_CORE,
PROTOCOL_COREPLUS,
PROTOCOL_LANMAN1,
PROTOCOL_LANMAN2,
PROTOCOL_NT1,
PROTOCOL_SMB2
};
int ms_fnmatch(const char *pattern, const char *string, enum protocol_types protocol);
/** a generic fnmatch function - uses for non-CIFS pattern matching */
int gen_fnmatch(const char *pattern, const char *string);
#endif
/* The following definitions come from lib/util/mutex.c */
#ifdef _SAMBA_BUILD_
/**
register a set of mutex/rwlock handlers.
Should only be called once in the execution of smbd.
*/
_PUBLIC_ bool register_mutex_handlers(const char *name, struct mutex_ops *ops);
#endif
/* The following definitions come from lib/util/idtree.c */
/**
initialise a idr tree. The context return value must be passed to
all subsequent idr calls. To destroy the idr tree use talloc_free()
on this context
*/
_PUBLIC_ struct idr_context *idr_init(TALLOC_CTX *mem_ctx);
/**
allocate the next available id, and assign 'ptr' into its slot.
you can retrieve later this pointer using idr_find()
*/
_PUBLIC_ int idr_get_new(struct idr_context *idp, void *ptr, int limit);
/**
allocate a new id, giving the first available value greater than or
equal to the given starting id
*/
_PUBLIC_ int idr_get_new_above(struct idr_context *idp, void *ptr, int starting_id, int limit);
/**
allocate a new id randomly in the given range
*/
_PUBLIC_ int idr_get_new_random(struct idr_context *idp, void *ptr, int limit);
/**
find a pointer value previously set with idr_get_new given an id
*/
_PUBLIC_ void *idr_find(struct idr_context *idp, int id);
/**
remove an id from the idr tree
*/
_PUBLIC_ int idr_remove(struct idr_context *idp, int id);
/* The following definitions come from lib/util/become_daemon.c */
#if _SAMBA_BUILD_ == 4
/**
Become a daemon, discarding the controlling terminal.
**/
_PUBLIC_ void become_daemon(bool fork);
#endif
/**
* Load a ini-style file.
*/
bool pm_process( const char *fileName,
bool (*sfunc)(const char *, void *),
bool (*pfunc)(const char *, const char *, void *),
void *userdata);
/**
* Add-on to talloc_get_type
*/
_PUBLIC_ void *talloc_check_name_abort(const void *ptr, const char *name);
#define talloc_get_type_abort(ptr, type) \
(type *)talloc_check_name_abort(ptr, #type)
bool unmap_file(void *start, size_t size);
#define CONST_DISCARD(type, ptr) ((type) ((void *) (ptr)))
#endif /* _SAMBA_UTIL_H_ */

2
ctdb/packaging/RPM/ctdb.spec.in
View File

@ -95,6 +95,7 @@ rm -rf $RPM_BUILD_ROOT
%{_sysconfdir}/ctdb/events.d/00.ctdb
%{_sysconfdir}/ctdb/events.d/01.reclock
%{_sysconfdir}/ctdb/events.d/10.interface
%{_sysconfdir}/ctdb/events.d/13.per_ip_routing
%{_sysconfdir}/ctdb/events.d/11.natgw
%{_sysconfdir}/ctdb/events.d/11.routing
%{_sysconfdir}/ctdb/events.d/20.multipathd
@ -107,6 +108,7 @@ rm -rf $RPM_BUILD_ROOT
%{_sysconfdir}/ctdb/events.d/70.iscsi
%{_sysconfdir}/ctdb/events.d/91.lvs
%{_sysconfdir}/ctdb/statd-callout
%{_sysconfdir}/ctdb/interface_modify.sh
%{_sbindir}/ctdbd
%{_bindir}/ctdb
%{_bindir}/smnotify

12
ctdb/server/ctdb_control.c
View File

@ -568,6 +568,18 @@ static int32_t ctdb_control_dispatch(struct ctdb_context *ctdb,
CHECK_CONTROL_DATA_SIZE(sizeof(uint32_t));
return ctdb_control_db_get_health(ctdb, indata, outdata);
case CTDB_CONTROL_GET_PUBLIC_IP_INFO:
CHECK_CONTROL_DATA_SIZE(sizeof(ctdb_sock_addr));
return ctdb_control_get_public_ip_info(ctdb, c, indata, outdata);
case CTDB_CONTROL_GET_IFACES:
CHECK_CONTROL_DATA_SIZE(0);
return ctdb_control_get_ifaces(ctdb, c, outdata);
case CTDB_CONTROL_SET_IFACE_LINK_STATE:
CHECK_CONTROL_DATA_SIZE(sizeof(struct ctdb_control_iface_info));
return ctdb_control_set_iface_link(ctdb, c, indata);
default:
DEBUG(DEBUG_CRIT,(__location__ " Unknown CTDB control opcode %u\n", opcode));
return -1;

6
ctdb/server/ctdb_daemon.c
View File

@ -781,6 +781,12 @@ int ctdb_start_daemon(struct ctdb_context *ctdb, bool do_fork, bool use_syslog)
ctdb_fatal(ctdb, "Failed to get initial freeze\n");
}
ret = ctdb_event_script(ctdb, CTDB_EVENT_INIT);
if (ret != 0) {
ctdb_fatal(ctdb, "Failed to run init event\n");
}
ctdb_run_notification_script(ctdb, "init");
/* now start accepting clients, only can do this once frozen */
fde = event_add_fd(ctdb->ev, ctdb, ctdb->daemon.sd,
EVENT_FD_READ|EVENT_FD_AUTOCLOSE,

2
ctdb/server/ctdb_monitor.c
View File

@ -75,7 +75,7 @@ static int ctdb_run_notification_script_child(struct ctdb_context *ctdb, const c
return ret;
}
static void ctdb_run_notification_script(struct ctdb_context *ctdb, const char *event)
void ctdb_run_notification_script(struct ctdb_context *ctdb, const char *event)
{
pid_t child;

212
ctdb/server/ctdb_recoverd.c
View File

@ -64,6 +64,7 @@ struct ctdb_recoverd {
TALLOC_CTX *ip_reallocate_ctx;
struct ip_reallocate_list *reallocate_callers;
TALLOC_CTX *ip_check_disable_ctx;
struct ctdb_control_get_ifaces *ifaces;
};
#define CONTROL_TIMEOUT() timeval_current_ofs(ctdb->tunable.recover_timeout, 0)
@ -1226,7 +1227,73 @@ static void reload_nodes_file(struct ctdb_context *ctdb)
ctdb_load_nodes_file(ctdb);
}
static int ctdb_reload_remote_public_ips(struct ctdb_context *ctdb,
struct ctdb_node_map *nodemap,
uint32_t *culprit)
{
int j;
int ret;
if (ctdb->num_nodes != nodemap->num) {
DEBUG(DEBUG_ERR, (__location__ " ctdb->num_nodes (%d) != nodemap->num (%d) invalid param\n",
ctdb->num_nodes, nodemap->num));
if (culprit) {
*culprit = ctdb->pnn;
}
return -1;
}
for (j=0; j<nodemap->num; j++) {
/* release any existing data */
if (ctdb->nodes[j]->known_public_ips) {
talloc_free(ctdb->nodes[j]->known_public_ips);
ctdb->nodes[j]->known_public_ips = NULL;
}
if (ctdb->nodes[j]->available_public_ips) {
talloc_free(ctdb->nodes[j]->available_public_ips);
ctdb->nodes[j]->available_public_ips = NULL;
}
if (nodemap->nodes[j].flags & NODE_FLAGS_INACTIVE) {
continue;
}
/* grab a new shiny list of public ips from the node */
ret = ctdb_ctrl_get_public_ips_flags(ctdb,
CONTROL_TIMEOUT(),
ctdb->nodes[j]->pnn,
ctdb->nodes,
0,
&ctdb->nodes[j]->known_public_ips);
if (ret != 0) {
DEBUG(DEBUG_ERR,("Failed to read known public ips from node : %u\n",
ctdb->nodes[j]->pnn));
if (culprit) {
*culprit = ctdb->nodes[j]->pnn;
}
return -1;
}
/* grab a new shiny list of public ips from the node */
ret = ctdb_ctrl_get_public_ips_flags(ctdb,
CONTROL_TIMEOUT(),
ctdb->nodes[j]->pnn,
ctdb->nodes,
CTDB_PUBLIC_IP_FLAGS_ONLY_AVAILABLE,
&ctdb->nodes[j]->available_public_ips);
if (ret != 0) {
DEBUG(DEBUG_ERR,("Failed to read available public ips from node : %u\n",
ctdb->nodes[j]->pnn));
if (culprit) {
*culprit = ctdb->nodes[j]->pnn;
}
return -1;
}
}
return 0;
}
/*
we are the recmaster, and recovery is needed - start a recovery run
*/
@ -1241,6 +1308,7 @@ static int do_recovery(struct ctdb_recoverd *rec,
TDB_DATA data;
uint32_t *nodes;
struct timeval start_time;
uint32_t culprit = (uint32_t)-1;
DEBUG(DEBUG_NOTICE, (__location__ " Starting do_recovery\n"));
@ -1500,6 +1568,12 @@ static int do_recovery(struct ctdb_recoverd *rec,
/*
tell nodes to takeover their public IPs
*/
ret = ctdb_reload_remote_public_ips(ctdb, nodemap, &culprit);
if (ret != 0) {
DEBUG(DEBUG_ERR,("Failed to read public ips from remote node %d\n",
culprit));
return -1;
}
rec->need_takeover_run = false;
ret = ctdb_takeover_run(ctdb, nodemap);
if (ret != 0) {
@ -1880,9 +1954,28 @@ static void process_ipreallocate_requests(struct ctdb_context *ctdb, struct ctdb
TDB_DATA result;
int32_t ret;
struct ip_reallocate_list *callers;
uint32_t culprit;
DEBUG(DEBUG_INFO, ("recovery master forced ip reallocation\n"));
ret = ctdb_takeover_run(ctdb, rec->nodemap);
/* update the list of public ips that a node can handle for
all connected nodes
*/
ret = ctdb_reload_remote_public_ips(ctdb, rec->nodemap, &culprit);
if (ret != 0) {
DEBUG(DEBUG_ERR,("Failed to read public ips from remote node %d\n",
culprit));
rec->need_takeover_run = true;
}
if (ret == 0) {
ret = ctdb_takeover_run(ctdb, rec->nodemap);
if (ret != 0) {
DEBUG(DEBUG_ERR,("Failed to read public ips from remote node %d\n",
culprit));
rec->need_takeover_run = true;
}
}
result.dsize = sizeof(int32_t);
result.dptr = (uint8_t *)&ret;
@ -2326,10 +2419,13 @@ static enum monitor_result verify_recmaster(struct ctdb_recoverd *rec, struct ct
static int verify_ip_allocation(struct ctdb_context *ctdb, struct ctdb_recoverd *rec, uint32_t pnn)
{
TALLOC_CTX *mem_ctx = talloc_new(NULL);
struct ctdb_control_get_ifaces *ifaces = NULL;
struct ctdb_all_public_ips *ips = NULL;
struct ctdb_uptime *uptime1 = NULL;
struct ctdb_uptime *uptime2 = NULL;
int ret, j;
bool need_iface_check = false;
bool need_takeover_run = false;
ret = ctdb_ctrl_uptime(ctdb, mem_ctx, CONTROL_TIMEOUT(),
CTDB_CURRENT_NODE, &uptime1);
@ -2339,6 +2435,30 @@ static int verify_ip_allocation(struct ctdb_context *ctdb, struct ctdb_recoverd
return -1;
}
/* read the interfaces from the local node */
ret = ctdb_ctrl_get_ifaces(ctdb, CONTROL_TIMEOUT(), CTDB_CURRENT_NODE, mem_ctx, &ifaces);
if (ret != 0) {
DEBUG(DEBUG_ERR, ("Unable to get interfaces from local node %u\n", pnn));
talloc_free(mem_ctx);
return -1;
}
if (!rec->ifaces) {
need_iface_check = true;
} else if (rec->ifaces->num != ifaces->num) {
need_iface_check = true;
} else if (memcmp(rec->ifaces, ifaces, talloc_get_size(ifaces)) != 0) {
need_iface_check = true;
}
if (need_iface_check) {
DEBUG(DEBUG_NOTICE, ("The interfaces status has changed on "
"local node %u - force takeover run\n",
pnn));
need_takeover_run = true;
}
/* read the ip allocation from the local node */
ret = ctdb_ctrl_get_public_ips(ctdb, CONTROL_TIMEOUT(), CTDB_CURRENT_NODE, mem_ctx, &ips);
if (ret != 0) {
@ -2380,6 +2500,9 @@ static int verify_ip_allocation(struct ctdb_context *ctdb, struct ctdb_recoverd
return 0;
}
talloc_free(rec->ifaces);
rec->ifaces = talloc_steal(rec, ifaces);
/* verify that we have the ip addresses we should have
and we dont have ones we shouldnt have.
if we find an inconsistency we set recmode to
@ -2389,43 +2512,35 @@ static int verify_ip_allocation(struct ctdb_context *ctdb, struct ctdb_recoverd
for (j=0; j<ips->num; j++) {
if (ips->ips[j].pnn == pnn) {
if (!ctdb_sys_have_ip(&ips->ips[j].addr)) {
struct takeover_run_reply rd;
TDB_DATA data;
DEBUG(DEBUG_CRIT,("Public address '%s' is missing and we should serve this ip\n",
ctdb_addr_to_str(&ips->ips[j].addr)));
rd.pnn = ctdb->pnn;
rd.srvid = 0;
data.dptr = (uint8_t *)&rd;
data.dsize = sizeof(rd);
ret = ctdb_send_message(ctdb, rec->recmaster, CTDB_SRVID_TAKEOVER_RUN, data);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " Failed to send ipreallocate to recmaster :%d\n", (int)rec->recmaster));
}
need_takeover_run = true;
}
} else {
if (ctdb_sys_have_ip(&ips->ips[j].addr)) {
struct takeover_run_reply rd;
TDB_DATA data;
DEBUG(DEBUG_CRIT,("We are still serving a public address '%s' that we should not be serving.\n",
ctdb_addr_to_str(&ips->ips[j].addr)));
rd.pnn = ctdb->pnn;
rd.srvid = 0;
data.dptr = (uint8_t *)&rd;
data.dsize = sizeof(rd);
ret = ctdb_send_message(ctdb, rec->recmaster, CTDB_SRVID_TAKEOVER_RUN, data);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " Failed to send ipreallocate to recmaster :%d\n", (int)rec->recmaster));
}
need_takeover_run = true;
}
}
}
if (need_takeover_run) {
struct takeover_run_reply rd;
TDB_DATA data;
DEBUG(DEBUG_CRIT,("Trigger takeoverrun\n"));
rd.pnn = ctdb->pnn;
rd.srvid = 0;
data.dptr = (uint8_t *)&rd;
data.dsize = sizeof(rd);
ret = ctdb_send_message(ctdb, rec->recmaster, CTDB_SRVID_TAKEOVER_RUN, data);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " Failed to send ipreallocate to recmaster :%d\n", (int)rec->recmaster));
}
}
talloc_free(mem_ctx);
return 0;
}
@ -3000,36 +3115,11 @@ again:
goto again;
}
/* update the list of public ips that a node can handle for
all connected nodes
*/
if (ctdb->num_nodes != nodemap->num) {
DEBUG(DEBUG_ERR, (__location__ " ctdb->num_nodes (%d) != nodemap->num (%d) reloading nodes file\n", ctdb->num_nodes, nodemap->num));
reload_nodes_file(ctdb);
goto again;
}
for (j=0; j<nodemap->num; j++) {
/* release any existing data */
if (ctdb->nodes[j]->public_ips) {
talloc_free(ctdb->nodes[j]->public_ips);
ctdb->nodes[j]->public_ips = NULL;
}
if (nodemap->nodes[j].flags & NODE_FLAGS_INACTIVE) {
continue;
}
/* grab a new shiny list of public ips from the node */
if (ctdb_ctrl_get_public_ips(ctdb, CONTROL_TIMEOUT(),
ctdb->nodes[j]->pnn,
ctdb->nodes,
&ctdb->nodes[j]->public_ips)) {
DEBUG(DEBUG_ERR,("Failed to read public ips from node : %u\n",
ctdb->nodes[j]->pnn));
goto again;
}
}
/* verify that all active nodes agree that we are the recmaster */
switch (verify_recmaster(rec, nodemap, pnn)) {
@ -3256,14 +3346,29 @@ again:
/* we might need to change who has what IP assigned */
if (rec->need_takeover_run) {
uint32_t culprit = (uint32_t)-1;
rec->need_takeover_run = false;
/* update the list of public ips that a node can handle for
all connected nodes
*/
ret = ctdb_reload_remote_public_ips(ctdb, nodemap, &culprit);
if (ret != 0) {
DEBUG(DEBUG_ERR,("Failed to read public ips from remote node %d\n",
culprit));
ctdb_set_culprit(rec, culprit);
do_recovery(rec, mem_ctx, pnn, nodemap, vnnmap);
goto again;
}
/* execute the "startrecovery" event script on all nodes */
ret = run_startrecovery_eventscript(rec, nodemap);
if (ret!=0) {
DEBUG(DEBUG_ERR, (__location__ " Unable to run the 'startrecovery' event on cluster\n"));
ctdb_set_culprit(rec, ctdb->pnn);
do_recovery(rec, mem_ctx, pnn, nodemap, vnnmap);
goto again;
}
ret = ctdb_takeover_run(ctdb, nodemap);
@ -3271,6 +3376,7 @@ again:
DEBUG(DEBUG_ERR, (__location__ " Unable to setup public takeover addresses - starting recovery\n"));
ctdb_set_culprit(rec, ctdb->pnn);
do_recovery(rec, mem_ctx, pnn, nodemap, vnnmap);
goto again;
}
/* execute the "recovered" event script on all nodes */

875
ctdb/server/ctdb_takeover.c
View File

File diff suppressed because it is too large Load Diff

17
ctdb/server/ctdbd.c
View File

@ -173,6 +173,7 @@ int main(int argc, const char *argv[])
talloc_enable_null_tracking();
ctdb_block_signal(SIGPIPE);
fault_setup("ctdbd");
ev = event_context_init(NULL);
@ -281,24 +282,14 @@ int main(int argc, const char *argv[])
}
if (options.single_public_ip) {
struct ctdb_vnn *svnn;
if (options.public_interface == NULL) {
DEBUG(DEBUG_ALERT,("--single_public_ip used but --public_interface is not specified. You must specify the public interface when using single public ip. Exiting\n"));
exit(10);
}
svnn = talloc_zero(ctdb, struct ctdb_vnn);
CTDB_NO_MEMORY(ctdb, svnn);
ctdb->single_ip_vnn = svnn;
svnn->iface = talloc_strdup(svnn, options.public_interface);
CTDB_NO_MEMORY(ctdb, svnn->iface);
if (parse_ip(options.single_public_ip,
svnn->iface,
0,
&svnn->public_address) == 0) {
ret = ctdb_set_single_public_ip(ctdb, options.public_interface,
options.single_public_ip);
if (ret != 0) {
DEBUG(DEBUG_ALERT,("Invalid --single-public-ip argument : %s . This is not a valid ip address. Exiting.\n", options.single_public_ip));
exit(10);
}

11
ctdb/server/eventscript.c
View File

@ -589,6 +589,7 @@ static bool check_options(enum ctdb_eventscript_call call, const char *options)
{
switch (call) {
/* These all take no arguments. */
case CTDB_EVENT_INIT:
case CTDB_EVENT_STARTUP:
case CTDB_EVENT_START_RECOVERY:
case CTDB_EVENT_RECOVERED:
@ -603,6 +604,9 @@ static bool check_options(enum ctdb_eventscript_call call, const char *options)
case CTDB_EVENT_RELEASE_IP:
return count_words(options) == 3;
case CTDB_EVENT_UPDATE_IP: /* old interface, new interface, IP address, netmask bits. */
return count_words(options) == 4;
default:
DEBUG(DEBUG_ERR,(__location__ "Unknown ctdb_eventscript_call %u\n", call));
return false;
@ -649,7 +653,12 @@ static int ctdb_event_script_callback_v(struct ctdb_context *ctdb,
/* we guarantee that only some specifically allowed event scripts are run
while in recovery */
const enum ctdb_eventscript_call allowed_calls[] = {
CTDB_EVENT_START_RECOVERY, CTDB_EVENT_SHUTDOWN, CTDB_EVENT_RELEASE_IP, CTDB_EVENT_STOPPED };
CTDB_EVENT_INIT,
CTDB_EVENT_START_RECOVERY,
CTDB_EVENT_SHUTDOWN,
CTDB_EVENT_RELEASE_IP,
CTDB_EVENT_STOPPED
};
int i;
for (i=0;i<ARRAY_SIZE(allowed_calls);i++) {
if (call == allowed_calls[i]) break;

11
ctdb/tests/events.d/00.test
View File

@ -16,9 +16,18 @@ case $cmd in
exit 0;
;;
init)
echo "ctdb init event"
exit 0;
;;
startup)
echo "ctdb startup event"
exit 0;
IFACES=`ctdb ifaces -Y | grep -v '^:Name:LinkStatus:References:'`
for I in $IFACES; do
IFACE=`echo -n "$I" | cut -d ':' -f2`
ctdb setifacelink $IFACE up
done
exit 0;
;;
takeip)

13
ctdb/tests/scripts/ctdb_test_functions.bash
View File

@ -239,7 +239,7 @@ sanity_check_output ()
sanity_check_ips ()
{
local ips="$1" # Output of "ctdb ip -n all"
local ips="$1" # list of "ip node" lines
echo "Sanity checking IPs..."
@ -259,9 +259,16 @@ sanity_check_ips ()
return 1
}
# This returns a list of "ip node" lines in $out
all_ips_on_node()
{
local node=$@
try_command_on_node $node "$CTDB ip -Y -n all | cut -d ':' -f1-3 | sed -e '1d' -e 's@^:@@' -e 's@:@ @g'"
}
select_test_node_and_ips ()
{
try_command_on_node 0 "$CTDB ip -n all | sed -e '1d'"
all_ips_on_node 0
# When selecting test_node we just want a node that has public
# IPs. This will work and is economically semi-random. :-)
@ -450,7 +457,7 @@ ips_are_on_nodeglob ()
local out
try_command_on_node 1 ctdb ip -n all
all_ips_on_node 1
while read ip pnn ; do
for check in $ips ; do

22
ctdb/tests/simple/11_ctdb_ip.sh
View File

@ -33,8 +33,20 @@ cluster_is_healthy
echo "Getting list of public IPs..."
try_command_on_node -v 1 $CTDB ip -n all
ips=$(echo "$out" | sed -e '1d')
colons=$(echo "$ips" | sed -e 's@^@:@' -e 's@$@:@' -e 's@ @:@')
ips=$(echo "$out" | sed \
-e '1d' \
-e 's@ node\[@ @' \
-e 's@\].*$@:@')
machineout1=":Public IP:Node:ActiveInterface:AvailableInterfaces:ConfiguredInterfaces:"
machineout2=$(echo "$out" | sed \
-e '1d' \
-e 's@^@:@' \
-e 's@ node\[@:@' \
-e 's@\] active\[@:@' \
-e 's@\] available\[@:@' \
-e 's@\] configured\[@:@' \
-e 's@\]$@:@')
machineout=`echo -e "$machineout1\n$machineout2"`
while read ip pnn ; do
try_command_on_node $pnn "ip addr show"
@ -48,14 +60,16 @@ done <<<"$ips" # bashism to avoid problem setting variable in pipeline.
[ "$testfailures" != 1 ] && echo "Looks good!"
cmd="$CTDB -Y ip -n all | sed -e '1d'"
cmd="$CTDB -Y ip -n all"
echo "Checking that \"$cmd\" produces expected output..."
try_command_on_node 1 "$cmd"
if [ "$out" = "$colons" ] ; then
if [ "$out" = "$machineout" ] ; then
echo "Yep, looks good!"
else
echo "Nope, it looks like this:"
echo "$out"
echo "Should be like this:"
echo "$machineout"
testfailures=1
fi

2
ctdb/tests/simple/16_ctdb_config_add_ip.sh
View File

@ -46,7 +46,7 @@ cluster_is_healthy
ctdb_restart_when_done
echo "Getting list of public IPs..."
try_command_on_node 0 "$CTDB ip -n all | sed -e '1d'"
all_ips_on_node 0
# When selecting test_node we just want a node that has public IPs.
# This will work and is economically semi-randomly. :-)

4
ctdb/tests/simple/17_ctdb_config_delete_ip.sh
View File

@ -21,7 +21,7 @@ Steps:
3. Delete one public IP address being be served by the node, using
'ctdb delip'.
4. Verify that the delete IP address is no longer listed using the
'ctdb ip' command.
all_ips_on_node helper function.
Expected results:
@ -42,7 +42,7 @@ cluster_is_healthy
ctdb_restart_when_done
echo "Getting list of public IPs..."
try_command_on_node -v 0 "$CTDB ip -n all | sed -e '1d'"
all_ips_on_node -v 0
# Select an IP/node to remove.
num_ips=$(echo "$out" | wc -l)

2
ctdb/tests/simple/23_ctdb_moveip.sh
View File

@ -50,7 +50,7 @@ if [ $num_nodes -lt 2 ] ; then
fi
echo "Getting list of public IPs..."
try_command_on_node -v 0 "$CTDB ip -n all | sed -e '1d'"
all_ips_on_node -v 0
sanity_check_ips "$out"

261
ctdb/tools/ctdb.c
View File

@ -502,18 +502,40 @@ static int control_status(struct ctdb_context *ctdb, int argc, const char **argv
}
if(options.machinereadable){
printf(":Node:IP:Disconnected:Banned:Disabled:Unhealthy:Stopped:\n");
printf(":Node:IP:Disconnected:Banned:Disabled:Unhealthy:Stopped:Inactive:PartiallyOnline:\n");
for(i=0;i<nodemap->num;i++){
int partially_online = 0;
int j;
if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
continue;
}
printf(":%d:%s:%d:%d:%d:%d:%d:\n", nodemap->nodes[i].pnn,
if (nodemap->nodes[i].flags == 0) {
struct ctdb_control_get_ifaces *ifaces;
ret = ctdb_ctrl_get_ifaces(ctdb, TIMELIMIT(),
nodemap->nodes[i].pnn,
ctdb, &ifaces);
if (ret == 0) {
for (j=0; j < ifaces->num; j++) {
if (ifaces->ifaces[j].link_state != 0) {
continue;
}
partially_online = 1;
break;
}
talloc_free(ifaces);
}
}
printf(":%d:%s:%d:%d:%d:%d:%d:%d:%d:\n", nodemap->nodes[i].pnn,
ctdb_addr_to_str(&nodemap->nodes[i].addr),
!!(nodemap->nodes[i].flags&NODE_FLAGS_DISCONNECTED),
!!(nodemap->nodes[i].flags&NODE_FLAGS_BANNED),
!!(nodemap->nodes[i].flags&NODE_FLAGS_PERMANENTLY_DISABLED),
!!(nodemap->nodes[i].flags&NODE_FLAGS_UNHEALTHY),
!!(nodemap->nodes[i].flags&NODE_FLAGS_STOPPED));
!!(nodemap->nodes[i].flags&NODE_FLAGS_STOPPED),
!!(nodemap->nodes[i].flags&NODE_FLAGS_INACTIVE),
partially_online);
}
return 0;
}
@ -530,6 +552,7 @@ static int control_status(struct ctdb_context *ctdb, int argc, const char **argv
{ NODE_FLAGS_UNHEALTHY, "UNHEALTHY" },
{ NODE_FLAGS_DELETED, "DELETED" },
{ NODE_FLAGS_STOPPED, "STOPPED" },
{ NODE_FLAGS_INACTIVE, "INACTIVE" },
};
char *flags_str = NULL;
int j;
@ -537,6 +560,23 @@ static int control_status(struct ctdb_context *ctdb, int argc, const char **argv
if (nodemap->nodes[i].flags & NODE_FLAGS_DELETED) {
continue;
}
if (nodemap->nodes[i].flags == 0) {
struct ctdb_control_get_ifaces *ifaces;
ret = ctdb_ctrl_get_ifaces(ctdb, TIMELIMIT(),
nodemap->nodes[i].pnn,
ctdb, &ifaces);
if (ret == 0) {
for (j=0; j < ifaces->num; j++) {
if (ifaces->ifaces[j].link_state != 0) {
continue;
}
flags_str = talloc_strdup(ctdb, "PARTIALLYONLINE");
break;
}
talloc_free(ifaces);
}
}
for (j=0;j<ARRAY_SIZE(flag_names);j++) {
if (nodemap->nodes[i].flags & flag_names[j].flag) {
if (flags_str == NULL) {
@ -1656,7 +1696,7 @@ static int control_ip(struct ctdb_context *ctdb, int argc, const char **argv)
}
if (options.machinereadable){
printf(":Public IP:Node:\n");
printf(":Public IP:Node:ActiveInterface:AvailableInterfaces:ConfiguredInterfaces:\n");
} else {
if (options.pnn == CTDB_BROADCAST_ALL) {
printf("Public IPs on ALL nodes\n");
@ -1666,11 +1706,215 @@ static int control_ip(struct ctdb_context *ctdb, int argc, const char **argv)
}
for (i=1;i<=ips->num;i++) {
if (options.machinereadable){
printf(":%s:%d:\n", ctdb_addr_to_str(&ips->ips[ips->num-i].addr), ips->ips[ips->num-i].pnn);
struct ctdb_control_public_ip_info *info = NULL;
int32_t pnn;
char *aciface = NULL;
char *avifaces = NULL;
char *cifaces = NULL;
if (options.pnn == CTDB_BROADCAST_ALL) {
pnn = ips->ips[ips->num-i].pnn;
} else {
printf("%s %d\n", ctdb_addr_to_str(&ips->ips[ips->num-i].addr), ips->ips[ips->num-i].pnn);
pnn = options.pnn;
}
if (pnn != -1) {
ret = ctdb_ctrl_get_public_ip_info(ctdb, TIMELIMIT(), pnn, ctdb,
&ips->ips[ips->num-i].addr, &info);
} else {
ret = -1;
}
if (ret == 0) {
int j;
for (j=0; j < info->num; j++) {
if (cifaces == NULL) {
cifaces = talloc_strdup(info,
info->ifaces[j].name);
} else {
cifaces = talloc_asprintf_append(cifaces,
",%s",
info->ifaces[j].name);
}
if (info->active_idx == j) {
aciface = info->ifaces[j].name;
}
if (info->ifaces[j].link_state == 0) {
continue;
}
if (avifaces == NULL) {
avifaces = talloc_strdup(info, info->ifaces[j].name);
} else {
avifaces = talloc_asprintf_append(avifaces,
",%s",
info->ifaces[j].name);
}
}
}
if (options.machinereadable){
printf(":%s:%d:%s:%s:%s:\n",
ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
ips->ips[ips->num-i].pnn,
aciface?aciface:"",
avifaces?avifaces:"",
cifaces?cifaces:"");
} else {
printf("%s node[%d] active[%s] available[%s] configured[%s]\n",
ctdb_addr_to_str(&ips->ips[ips->num-i].addr),
ips->ips[ips->num-i].pnn,
aciface?aciface:"",
avifaces?avifaces:"",
cifaces?cifaces:"");
}
talloc_free(info);
}
talloc_free(tmp_ctx);
return 0;
}
/*
public ip info
*/
static int control_ipinfo(struct ctdb_context *ctdb, int argc, const char **argv)
{
int i, ret;
ctdb_sock_addr addr;
TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
struct ctdb_control_public_ip_info *info;
if (argc != 1) {
talloc_free(tmp_ctx);
usage();
}
if (parse_ip(argv[0], NULL, 0, &addr) == 0) {
DEBUG(DEBUG_ERR,("Wrongly formed ip address '%s'\n", argv[0]));
return -1;
}
/* read the public ip info from this node */
ret = ctdb_ctrl_get_public_ip_info(ctdb, TIMELIMIT(), options.pnn,
tmp_ctx, &addr, &info);
if (ret != 0) {
DEBUG(DEBUG_ERR, ("Unable to get public ip[%s]info from node %u\n",
argv[0], options.pnn));
talloc_free(tmp_ctx);
return ret;
}
printf("Public IP[%s] info on node %u\n",
ctdb_addr_to_str(&info->ip.addr),
options.pnn);
printf("IP:%s\nCurrentNode:%d\nNumInterfaces:%u\n",
ctdb_addr_to_str(&info->ip.addr),
info->ip.pnn, info->num);
for (i=0; i<info->num; i++) {
info->ifaces[i].name[CTDB_IFACE_SIZE] = '\0';
printf("Interface[%u]: Name:%s Link:%s References:%u%s\n",
i+1, info->ifaces[i].name,
info->ifaces[i].link_state?"up":"down",
(unsigned int)info->ifaces[i].references,
(i==info->active_idx)?" (active)":"");
}
talloc_free(tmp_ctx);
return 0;
}
/*
display interfaces status
*/
static int control_ifaces(struct ctdb_context *ctdb, int argc, const char **argv)
{
int i, ret;
TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
struct ctdb_control_get_ifaces *ifaces;
/* read the public ip list from this node */
ret = ctdb_ctrl_get_ifaces(ctdb, TIMELIMIT(), options.pnn,
tmp_ctx, &ifaces);
if (ret != 0) {
DEBUG(DEBUG_ERR, ("Unable to get interfaces from node %u\n",
options.pnn));
talloc_free(tmp_ctx);
return ret;
}
if (options.machinereadable){
printf(":Name:LinkStatus:References:\n");
} else {
printf("Interfaces on node %u\n", options.pnn);
}
for (i=0; i<ifaces->num; i++) {
if (options.machinereadable){
printf(":%s:%s:%u\n",
ifaces->ifaces[i].name,
ifaces->ifaces[i].link_state?"1":"0",
(unsigned int)ifaces->ifaces[i].references);
} else {
printf("name:%s link:%s references:%u\n",
ifaces->ifaces[i].name,
ifaces->ifaces[i].link_state?"up":"down",
(unsigned int)ifaces->ifaces[i].references);
}
}
talloc_free(tmp_ctx);
return 0;
}
/*
set link status of an interface
*/
static int control_setifacelink(struct ctdb_context *ctdb, int argc, const char **argv)
{
int ret;
TALLOC_CTX *tmp_ctx = talloc_new(ctdb);
struct ctdb_control_iface_info info;
ZERO_STRUCT(info);
if (argc != 2) {
usage();
}
if (strlen(argv[0]) > CTDB_IFACE_SIZE) {
DEBUG(DEBUG_ERR, ("interfaces name '%s' too long\n",
argv[0]));
talloc_free(tmp_ctx);
return -1;
}
strcpy(info.name, argv[0]);
if (strcmp(argv[1], "up") == 0) {
info.link_state = 1;
} else if (strcmp(argv[1], "down") == 0) {
info.link_state = 0;
} else {
DEBUG(DEBUG_ERR, ("link state invalid '%s' should be 'up' or 'down'\n",
argv[1]));
talloc_free(tmp_ctx);
return -1;
}
/* read the public ip list from this node */
ret = ctdb_ctrl_set_iface_link(ctdb, TIMELIMIT(), options.pnn,
tmp_ctx, &info);
if (ret != 0) {
DEBUG(DEBUG_ERR, ("Unable to set link state for interfaces %s node %u\n",
argv[0], options.pnn));
talloc_free(tmp_ctx);
return ret;
}
talloc_free(tmp_ctx);
@ -4002,6 +4246,9 @@ static const struct {
{ "statistics", control_statistics, false, false, "show statistics" },
{ "statisticsreset", control_statistics_reset, true, false, "reset statistics"},
{ "ip", control_ip, false, false, "show which public ip's that ctdb manages" },
{ "ipinfo", control_ipinfo, true, false, "show details about a public ip that ctdb manages", "<ip>" },
{ "ifaces", control_ifaces, true, false, "show which interfaces that ctdb manages" },
{ "setifacelink", control_setifacelink, true, false, "set interface link status", "<iface> <status>" },
{ "process-exists", control_process_exists, true, false, "check if a process exists on a node", "<pid>"},
{ "getdbmap", control_getdbmap, true, false, "show the database map" },
{ "getdbstatus", control_getdbstatus, true, false, "show the status of a database", "<dbname>" },