diff options
67 files changed, 1991 insertions, 172 deletions
diff --git a/ctdb/Makefile.in b/ctdb/Makefile.in index 80b0a02f55..7ec430c164 100755 --- a/ctdb/Makefile.in +++ b/ctdb/Makefile.in @@ -29,6 +29,10 @@ POPT_LIBS = @POPT_LIBS@ POPT_CFLAGS = @POPT_CFLAGS@ POPT_OBJ = @POPT_OBJ@ +PMDA_LIBS = -lpcp -lpcp_pmda +PMDA_INSTALL = @CTDB_PMDA_INSTALL@ +PMDA_DEST_DIR = /var/lib/pcp/pmdas + CFLAGS=-g -I$(srcdir)/include -Iinclude -Ilib -Ilib/util -I$(srcdir) \ -I@tallocdir@ -I@tdbdir@/include -I@libreplacedir@ \ -DVARDIR=\"$(localstatedir)\" -DETCDIR=\"$(etcdir)\" \ @@ -75,7 +79,8 @@ TEST_BINS=tests/bin/ctdb_bench tests/bin/ctdb_fetch tests/bin/ctdb_fetch_one \ tests/bin/ctdb_takeover_tests tests/bin/ctdb_update_record \ @INFINIBAND_BINS@ -BINS = bin/ctdb @CTDB_SCSI_IO@ bin/smnotify bin/ping_pong bin/ltdbtool +BINS = bin/ctdb @CTDB_SCSI_IO@ bin/smnotify bin/ping_pong bin/ltdbtool @CTDB_PMDA@ + SBINS = bin/ctdbd DIRS = lib bin tests/bin @@ -152,6 +157,9 @@ bin/ping_pong: utils/ping_pong/ping_pong.o @echo Linking $@ @$(CC) $(CFLAGS) -o $@ utils/ping_pong/ping_pong.o +bin/pmdactdb: $(CTDB_CLIENT_OBJ) utils/pmda/pmda_ctdb.o + @echo Linking $@ + @$(CC) $(CFLAGS) -o $@ utils/pmda/pmda_ctdb.o $(CTDB_CLIENT_OBJ) $(LIB_FLAGS) $(PMDA_LIBS) tests/bin/rb_test: $(CTDB_CLIENT_OBJ) tests/src/rb_test.o @echo Linking $@ @@ -171,7 +179,7 @@ tests/bin/ctdb_fetch_one: $(CTDB_CLIENT_OBJ) tests/src/ctdb_fetch_one.o tests/bin/ctdb_fetch_lock_once: libctdb/libctdb.a tests/src/ctdb_fetch_lock_once.o @echo Linking $@ - @$(CC) $(CFLAGS) -o $@ tests/src/ctdb_fetch_lock_once.o libctdb/libctdb.a -ltdb $(LIB_FLAGS) + @$(CC) $(CFLAGS) -o $@ tests/src/ctdb_fetch_lock_once.o @TDB_OBJ@ @POPT_OBJ@ libctdb/libctdb.a $(LIB_FLAGS) tests/bin/ctdb_fetch_readonly_once: $(CTDB_CLIENT_OBJ) tests/src/ctdb_fetch_readonly_once.o @echo Linking $@ @@ -242,7 +250,7 @@ distclean: clean rm -f config.log config.status config.cache config.h rm -f Makefile -install: all +install: all $(PMDA_INSTALL) mkdir -p $(DESTDIR)$(libdir)/pkgconfig mkdir -p $(DESTDIR)$(bindir) mkdir -p $(DESTDIR)$(sbindir) @@ -295,6 +303,12 @@ install: all if [ ! -f $(DESTDIR)$(etcdir)/ctdb/notify.sh ];then ${INSTALLCMD} -m 755 config/notify.sh $(DESTDIR)$(etcdir)/ctdb; fi if [ ! -f $(DESTDIR)$(etcdir)/ctdb/ctdb-crash-cleanup.sh ];then ${INSTALLCMD} -m 755 config/ctdb-crash-cleanup.sh $(DESTDIR)$(etcdir)/ctdb; fi +install_pmda: + $(INSTALLCMD) -m 755 -d $(PMDA_DEST_DIR) + $(INSTALLCMD) -m 755 pmda/Install pmda/Remove $(PMDA_DEST_DIR) + $(INSTALLCMD) -m 644 pmda/pmns pmda/domain.h pmda/help pmda/README $(PMDA_DEST_DIR) + $(INSTALLCMD) -m 755 bin/pmdactdb $(PMDA_DEST_DIR) + test: all tests/run_tests.sh diff --git a/ctdb/client/ctdb_client.c b/ctdb/client/ctdb_client.c index b3061a855c..89eeb4836a 100644 --- a/ctdb/client/ctdb_client.c +++ b/ctdb/client/ctdb_client.c @@ -187,7 +187,7 @@ static void ctdb_client_reply_control(struct ctdb_context *ctdb, struct ctdb_req /* this is called in the client, when data comes in from the daemon */ -static void ctdb_client_read_cb(uint8_t *data, size_t cnt, void *args) +void ctdb_client_read_cb(uint8_t *data, size_t cnt, void *args) { struct ctdb_context *ctdb = talloc_get_type(args, struct ctdb_context); struct ctdb_req_header *hdr = (struct ctdb_req_header *)data; @@ -1944,7 +1944,11 @@ static int ctdb_fetch_with_header_func(struct ctdb_call_info *call) /* attach to a specific database - client call */ -struct ctdb_db_context *ctdb_attach(struct ctdb_context *ctdb, const char *name, bool persistent, uint32_t tdb_flags) +struct ctdb_db_context *ctdb_attach(struct ctdb_context *ctdb, + struct timeval timeout, + const char *name, + bool persistent, + uint32_t tdb_flags) { struct ctdb_db_context *ctdb_db; TDB_DATA data; @@ -1979,7 +1983,7 @@ struct ctdb_db_context *ctdb_attach(struct ctdb_context *ctdb, const char *name, ctdb_db->db_id = *(uint32_t *)data.dptr; talloc_free(data.dptr); - ret = ctdb_ctrl_getdbpath(ctdb, timeval_current_ofs(2, 0), CTDB_CURRENT_NODE, ctdb_db->db_id, ctdb_db, &ctdb_db->db_path); + ret = ctdb_ctrl_getdbpath(ctdb, timeout, CTDB_CURRENT_NODE, ctdb_db->db_id, ctdb_db, &ctdb_db->db_path); if (ret != 0) { DEBUG(DEBUG_ERR,("Failed to get dbpath for database '%s'\n", name)); talloc_free(ctdb_db); diff --git a/ctdb/common/cmdline.c b/ctdb/common/cmdline.c index 145a13a81b..5c00763ed7 100644 --- a/ctdb/common/cmdline.c +++ b/ctdb/common/cmdline.c @@ -111,7 +111,8 @@ struct ctdb_context *ctdb_cmdline_init(struct event_context *ev) /* startup a client only ctdb context */ -struct ctdb_context *ctdb_cmdline_client(struct event_context *ev) +struct ctdb_context *ctdb_cmdline_client(struct tevent_context *ev, + struct timeval req_timeout) { struct ctdb_context *ctdb; char *socket_name; @@ -152,7 +153,7 @@ struct ctdb_context *ctdb_cmdline_client(struct event_context *ev) } /* get our pnn */ - ctdb->pnn = ctdb_ctrl_getpnn(ctdb, timeval_current_ofs(3, 0), CTDB_CURRENT_NODE); + ctdb->pnn = ctdb_ctrl_getpnn(ctdb, req_timeout, CTDB_CURRENT_NODE); if (ctdb->pnn == (uint32_t)-1) { DEBUG(DEBUG_CRIT,(__location__ " Failed to get ctdb pnn\n")); talloc_free(ctdb); diff --git a/ctdb/common/system_aix.c b/ctdb/common/system_aix.c index 5fe54130d4..1404a82901 100644 --- a/ctdb/common/system_aix.c +++ b/ctdb/common/system_aix.c @@ -357,4 +357,8 @@ int ctdb_sys_read_tcp_packet(int s, void *private_data, } +bool ctdb_sys_check_iface_exists(const char *iface) +{ + return true; +} diff --git a/ctdb/common/system_common.c b/ctdb/common/system_common.c index f28045f95e..3fe90e6ef3 100644 --- a/ctdb/common/system_common.c +++ b/ctdb/common/system_common.c @@ -73,3 +73,5 @@ bool ctdb_sys_have_ip(ctdb_sock_addr *_addr) close(s); return ret == 0; } + + diff --git a/ctdb/common/system_linux.c b/ctdb/common/system_linux.c index a498ab2d68..2dcdffb3ec 100644 --- a/ctdb/common/system_linux.c +++ b/ctdb/common/system_linux.c @@ -537,3 +537,25 @@ int ctdb_sys_read_tcp_packet(int s, void *private_data, } +bool ctdb_sys_check_iface_exists(const char *iface) +{ + int s; + struct ifreq ifr; + + s = socket(PF_PACKET, SOCK_RAW, 0); + if (s == -1){ + /* We dont know if the interface exists, so assume yes */ + DEBUG(DEBUG_CRIT,(__location__ " failed to open raw socket\n")); + return true; + } + + strncpy(ifr.ifr_name, iface, sizeof(ifr.ifr_name)); + if (ioctl(s, SIOCGIFINDEX, &ifr) < 0 && errno == ENODEV) { + DEBUG(DEBUG_CRIT,(__location__ " interface '%s' not found\n", iface)); + close(s); + return false; + } + close(s); + + return true; +} diff --git a/ctdb/config/ctdb.sysconfig b/ctdb/config/ctdb.sysconfig index 5115094673..163a0dd6ab 100644 --- a/ctdb/config/ctdb.sysconfig +++ b/ctdb/config/ctdb.sysconfig @@ -295,6 +295,10 @@ CTDB_DEBUGLEVEL=ERR # above 90% # CTDB_CHECK_FS_USE="/:90 /var:90" +# Should CTDB automatically start and stop services when it is told to +# newly manage or no longer manage them? +CTDB_SERVICE_AUTOSTARTSTOP=yes + # # # set any default tuning options for ctdb diff --git a/ctdb/config/events.d/00.ctdb b/ctdb/config/events.d/00.ctdb index 321ba9c708..2a48afb8e2 100755 --- a/ctdb/config/events.d/00.ctdb +++ b/ctdb/config/events.d/00.ctdb @@ -35,6 +35,8 @@ update_config_from_tdb() { fi } +ctdb_check_args "$@" + case "$1" in init) # make sure we have a blank state directory for the scripts to work with diff --git a/ctdb/config/events.d/10.interface b/ctdb/config/events.d/10.interface index f75e6afea4..356ee08245 100755 --- a/ctdb/config/events.d/10.interface +++ b/ctdb/config/events.d/10.interface @@ -129,6 +129,8 @@ monitor_interfaces() return 1 } +ctdb_check_args "$@" + case "$1" in ############################# # called when ctdbd starts up @@ -158,10 +160,6 @@ case "$1" in ################################################ # called when ctdbd wants to claim an IP address takeip) - if [ $# != 4 ]; then - echo "ERROR: must supply interface, IP and maskbits" - exit 1 - fi iface=$2 ip=$3 maskbits=$4 @@ -181,11 +179,6 @@ case "$1" in ################################################## # called when ctdbd wants to release an IP address releaseip) - if [ $# != 4 ]; then - echo "ERROR: must supply interface, IP and maskbits" - exit 1 - fi - # releasing an IP is a bit more complex than it seems. Once the IP # is released, any open tcp connections to that IP on this host will end # up being stuck. Some of them (such as NFS connections) will be unkillable @@ -220,11 +213,6 @@ case "$1" in ################################################## # called when ctdbd wants to update an IP address updateip) - if [ $# != 5 ]; then - echo "ERROR: 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 diff --git a/ctdb/config/events.d/13.per_ip_routing b/ctdb/config/events.d/13.per_ip_routing index ff4dbdb8c3..d8a064493e 100755 --- a/ctdb/config/events.d/13.per_ip_routing +++ b/ctdb/config/events.d/13.per_ip_routing @@ -326,6 +326,8 @@ setup_per_ip_routing() return $?; } +ctdb_check_args "$@" + case "$1" in ############################# # called when ctdbd starts up @@ -368,10 +370,6 @@ case "$1" in ################################################ # 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 @@ -410,10 +408,6 @@ case "$1" in ################################################ # 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 @@ -455,11 +449,6 @@ case "$1" in ################################################## # 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 diff --git a/ctdb/config/events.d/60.nfs b/ctdb/config/events.d/60.nfs index 87a91df5b5..e8ac61fe43 100755 --- a/ctdb/config/events.d/60.nfs +++ b/ctdb/config/events.d/60.nfs @@ -70,6 +70,7 @@ case "$1" in [ "$CTDB_NFS_SKIP_SHARE_CHECK" = "yes" ] || { exportfs -v | grep '^/' | sed -r -e 's@[[:space:]]+[^[:space:]()]+\([^[:space:]()]+\)$@@' | + sort -u | ctdb_check_directories } || exit $? diff --git a/ctdb/config/functions b/ctdb/config/functions index f65690359b..5ca7d8e693 100755 --- a/ctdb/config/functions +++ b/ctdb/config/functions @@ -83,6 +83,26 @@ debug () } ############################################################## +# check number of args for different events +ctdb_check_args () +{ + case "$1" in + takeip|releaseip) + if [ $# != 4 ]; then + echo "ERROR: must supply interface, IP and maskbits" + exit 1 + fi + ;; + updateip) + if [ $# != 5 ]; then + echo "ERROR: must supply old interface, new interface, IP and maskbits" + exit 1 + fi + ;; + esac +} + +############################################################## # determine on what type of system (init style) we are running detect_init_style() { # only do detection if not already set: @@ -358,6 +378,7 @@ ctdb_check_tcp_ports() ctdb_check_tcp_ports_$_c "$@" case "$?" in 0) + _ctdb_check_tcp_common rm -f "$_ctdb_service_started_file" return 0 ;; @@ -466,7 +487,8 @@ ctdb_check_tcp_ports_ctdb () for _p ; do # process each function argument (port) _cmd="ctdb checktcpport $_p" _out=$($_cmd 2>&1) - case "$?" in + _ret=$? + case "$_ret" in 0) ctdb_check_tcp_ports_debug="\"$_cmd\" was able to bind to port" return 1 @@ -476,7 +498,7 @@ ctdb_check_tcp_ports_ctdb () continue ;; *) - ctdb_check_tcp_ports_debug="$_cmd (exited with $?) with output: + ctdb_check_tcp_ports_debug="$_cmd (exited with $_ret) with output: $_out" # assume not implemented return 127 @@ -992,7 +1014,7 @@ ctdb_service_unset_reconfigure () ctdb_service_reconfigure () { - echo "Reconfiguring service \"$service_name\"..." + echo "Reconfiguring service \"$@\"..." ctdb_service_unset_reconfigure "$@" service_reconfigure "$@" || return $? ctdb_counter_init "$@" @@ -1004,28 +1026,129 @@ service_reconfigure () service "${1:-$service_name}" restart } -ctdb_service_check_reconfigure () +ctdb_reconfigure_try_lock () { - # Only do this for certain events. - case "$event_name" in - monitor|ipreallocated) : ;; - *) return 0 + + _ctdb_service_reconfigure_common "$@" + _lock="${_d}/reconfigure_lock" + touch "$_lock" + + ( + flock 0 + # This is overkill but will work if we need to extend this to + # allow certain events to run multiple times in parallel + # (e.g. takeip) and write multiple PIDs to the file. + read _locker_event + if [ -n "$_locker_event" ] ; then + while read _pid ; do + if [ -n "$_pid" -a "$_pid" != $$ ] && \ + kill -0 "$_pid" 2>/dev/null ; then + exit 1 + fi + done + fi + + printf "%s\n%s\n" "$event_name" $$ >"$_lock" + exit 0 + ) <"$_lock" +} + +ctdb_replay_monitor_status () +{ + echo "Replaying previous status for this script due to reconfigure..." + # Leading colon (':') is missing in some versions... + _out=$(ctdb scriptstatus -Y | grep -E "^:?monitor:${script_name}:") + # Output looks like this: + # :monitor:60.nfs:1:ERROR:1314764004.030861:1314764004.035514:foo bar: + # This is the cheapest way of getting fields in the middle. + set -- $(IFS=":" ; echo $_out) + _code="$3" + _status="$4" + # The error output field can include colons so we'll try to + # preserve them. The weak checking at the beginning tries to make + # this work for both broken (no leading ':') and fixed output. + _out="${_out%:}" + _err_out="${_out#*monitor:${script_name}:*:*:*:*:}" + case "$_status" in + OK) : ;; # Do nothing special. + TIMEDOUT) + # Recast this as an error, since we can't exit with the + # correct negative number. + _code=1 + _err_out="[Replay of TIMEDOUT scriptstatus - note incorrect return code.] ${_err_out}" + ;; + DISABLED) + # Recast this as an OK, since we can't exit with the + # correct negative number. + _code=0 + _err_out="[Replay of DISABLED scriptstatus - note incorrect return code.] ${_err_out}" + ;; + *) : ;; # Must be ERROR, do nothing special. esac + echo "$_err_out" + exit $_code +} - if ctdb_service_needs_reconfigure "$@" ; then - ctdb_service_reconfigure "$@" +ctdb_service_check_reconfigure () +{ + [ -n "$1" ] || set -- "$service_name" - # Fall through to non-monitor events. - [ "$event_name" = "monitor" ] || return 0 + # We only care about some events in this function. For others we + # return now. + case "$event_name" in + monitor|ipreallocated|reconfigure) : ;; + *) return 0 ;; + esac - # We don't want to proceed with the rest of the monitor event - # here, so we exit. However, if we exit 0 then, if the - # service was previously broken, we might return a false - # positive. So we simply retrieve the status of this script - # from the previous monitor loop and exit with that status. - ctdb scriptstatus | \ - grep -q -E "^${script_name}[[:space:]]+Status:OK[[:space:]]" - exit $? + if ctdb_reconfigure_try_lock "$@" ; then + # No events covered by this function are running, so proceed + # with gay abandon. + case "$event_name" in + reconfigure) + (ctdb_service_reconfigure "$@") + exit $? + ;; + ipreallocated) + if ctdb_service_needs_reconfigure "$@" ; then + ctdb_service_reconfigure "$@" + fi + ;; + monitor) + if ctdb_service_needs_reconfigure "$@" ; then + ctdb_service_reconfigure "$@" + # Given that the reconfigure might not have + # resulted in the service being stable yet, we + # replay the previous status since that's the best + # information we have. + ctdb_replay_monitor_status + fi + ;; + esac + else + # Somebody else is running an event we don't want to collide + # with. We proceed with caution. + case "$event_name" in + reconfigure) + # Tell whoever called us to retry. + exit 2 + ;; + ipreallocated) + # Defer any scheduled reconfigure and just run the + # rest of the ipreallocated event, as per the + # eventscript. There's an assumption here that the + # event doesn't depend on any scheduled reconfigure. + # This is true in the current code. + return 0 + ;; + monitor) + # There is most likely a reconfigure in progress so + # the service is possibly unstable. As above, we + # defer any scheduled reconfigured. We also replay + # the previous monitor status since that's the best + # information we have. + ctdb_replay_monitor_status + ;; + esac fi } @@ -1072,6 +1195,9 @@ is_ctdb_managed_service () ctdb_start_stop_service () { + # Do nothing unless configured to... + [ "$CTDB_SERVICE_AUTOSTARTSTOP" = "yes" ] || return 0 + _service_name="${1:-${service_name}}" [ "$event_name" = "monitor" ] || return 0 diff --git a/ctdb/configure.ac b/ctdb/configure.ac index cc3a1143da..adcc5dadad 100644 --- a/ctdb/configure.ac +++ b/ctdb/configure.ac @@ -64,6 +64,15 @@ m4_include(lib/util/signal.m4) m4_include(lib/util/fault.m4) AC_CHECK_HEADERS(sched.h) +AC_CHECK_HEADERS(pcp/pmapi.h pcp/impl.h pcp/pmda.h, [], [], +[[#ifdef HAVE_PCP_PMAPI_H +# include <pcp/pmapi.h> +#endif +#ifdef HAVE_PCP_IMPL_H +# include <pcp/impl.h> +#endif +]]) + AC_CHECK_FUNCS(sched_setscheduler) AC_CHECK_FUNCS(mlockall) @@ -77,9 +86,20 @@ if test x"$ctdb_cv_HAVE_SOCK_SIN_LEN" = x"yes"; then AC_DEFINE(HAVE_SOCK_SIN_LEN,1,[Whether the sockaddr_in struct has a sin_len property]) fi +if test x"$ac_cv_header_pcp_pmda_h" = x"yes"; then + CTDB_PMDA=bin/pmdactdb + CTDB_PMDA_INSTALL=install_pmda +else + CTDB_PMDA= + CTDB_PMDA_INSTALL= +fi + + AC_SUBST(EXTRA_OBJ) AC_SUBST(CTDB_SYSTEM_OBJ) AC_SUBST(CTDB_SCSI_IO) AC_SUBST(CTDB_PCAP_LDFLAGS) +AC_SUBST(CTDB_PMDA) +AC_SUBST(CTDB_PMDA_INSTALL) AC_OUTPUT(Makefile ctdb.pc) diff --git a/ctdb/include/ctdb.h b/ctdb/include/ctdb.h index 484902fb83..4cdbc0f76a 100644 --- a/ctdb/include/ctdb.h +++ b/ctdb/include/ctdb.h @@ -116,6 +116,45 @@ void ctdb_disconnect(struct ctdb_connection *ctdb); ***/ /** + * ctdb_num_active - get the number of active commands + * @ctdb: the ctdb_connection from ctdb_connect. + * + * This command can be used to find the number of active commands we have + * issued. An active command is a command we have queued, or sent + * to the ctdb daemon but which we have not yet received a reply to. + * + * See Also: + * ctdb_num_in_flight(), ctdb_num_out_queue() + */ +int ctdb_num_active(struct ctdb_connection *ctdb); + +/** + * ctdb_num_in_flight - get the number of commands in flight. + * @ctdb: the ctdb_connection from ctdb_connect. + * + * This command can be used to find the number of commands we have + * sent to the ctdb daemon to which we have not yet received/processed + * the reply. + * + * See Also: + * ctdb_num_out_queue(), ctdb_num_active() + */ +int ctdb_num_in_flight(struct ctdb_connection *ctdb); + +/** + * ctdb_num_out_queue - get the number of commands in the out queue + * @ctdb: the ctdb_connection from ctdb_connect. + * + * This command can be used to find the number of commands we have + * queued for delivery to the ctdb daemon but have not yet been + * written to the domain socket. + * + * See Also: + * ctdb_num_in_flight(), ctdb_num_active() + */ +int ctdb_num_out_queue(struct ctdb_connection *ctdb); + +/** * ctdb_get_fd - get the filedescriptor to select/poll on * @ctdb: the ctdb_connection from ctdb_connect. * diff --git a/ctdb/include/ctdb_client.h b/ctdb/include/ctdb_client.h index ad31d24b31..01219f341b 100644 --- a/ctdb/include/ctdb_client.h +++ b/ctdb/include/ctdb_client.h @@ -105,7 +105,12 @@ int ctdb_start_daemon(struct ctdb_context *ctdb, bool do_fork, bool use_syslog, /* attach to a ctdb database */ -struct ctdb_db_context *ctdb_attach(struct ctdb_context *ctdb, const char *name, bool persistent, uint32_t tdb_flags); +struct ctdb_db_context *ctdb_attach(struct ctdb_context *ctdb, + struct timeval timeout, + const char *name, + bool persistent, + uint32_t tdb_flags); + /* find an attached ctdb_db handle given a name @@ -190,7 +195,8 @@ int ctdb_register_message_handler(struct ctdb_context *ctdb, struct ctdb_db_context *find_ctdb_db(struct ctdb_context *ctdb, uint32_t id); -struct ctdb_context *ctdb_cmdline_client(struct tevent_context *ev); +struct ctdb_context *ctdb_cmdline_client(struct tevent_context *ev, + struct timeval req_timeout); struct ctdb_statistics; int ctdb_ctrl_statistics(struct ctdb_context *ctdb, uint32_t destnode, struct ctdb_statistics *status); diff --git a/ctdb/include/ctdb_private.h b/ctdb/include/ctdb_private.h index 6a8ed187ef..b093bf09ac 100644 --- a/ctdb/include/ctdb_private.h +++ b/ctdb/include/ctdb_private.h @@ -121,6 +121,7 @@ struct ctdb_tunable { uint32_t deferred_attach_timeout; uint32_t vacuum_fast_path_count; uint32_t lcp2_public_ip_assignment; + uint32_t allow_client_db_attach; }; /* @@ -413,6 +414,7 @@ struct ctdb_context { uint32_t recovery_mode; TALLOC_CTX *tickle_update_context; TALLOC_CTX *keepalive_ctx; + TALLOC_CTX *check_public_ifaces_ctx; struct ctdb_tunable tunable; enum ctdb_freeze_mode freeze_mode[NUM_DB_PRIORITIES+1]; struct ctdb_freeze_handle *freeze_handles[NUM_DB_PRIORITIES+1]; @@ -802,6 +804,7 @@ int ctdb_call_local(struct ctdb_db_context *ctdb_db, struct ctdb_call *call, void ctdb_recv_raw_pkt(void *p, uint8_t *data, uint32_t length); int ctdb_socket_connect(struct ctdb_context *ctdb); +void ctdb_client_read_cb(uint8_t *data, size_t cnt, void *args); #define CTDB_BAD_REQID ((uint32_t)-1) uint32_t ctdb_reqid_new(struct ctdb_context *ctdb, void *state); @@ -1116,6 +1119,7 @@ int ctdb_ctrl_set_iface_link(struct ctdb_context *ctdb, uint32_t uint16_checksum(uint16_t *data, size_t n); int ctdb_sys_send_arp(const ctdb_sock_addr *addr, const char *iface); bool ctdb_sys_have_ip(ctdb_sock_addr *addr); +bool ctdb_sys_check_iface_exists(const char *iface); int ctdb_sys_send_tcp(const ctdb_sock_addr *dest, const ctdb_sock_addr *src, uint32_t seq, uint32_t ack, int rst); @@ -1358,6 +1362,7 @@ int32_t ctdb_control_clear_log(struct ctdb_context *ctdb); struct ctdb_log_state *ctdb_fork_with_logging(TALLOC_CTX *mem_ctx, struct ctdb_context *ctdb, + const char *log_prefix, void (*logfn)(const char *, uint16_t, void *), void *logfn_private, pid_t *pid); diff --git a/ctdb/lib/tevent/tevent_util.c b/ctdb/lib/tevent/tevent_util.c index 28c9da346a..54c1787400 100644 --- a/ctdb/lib/tevent/tevent_util.c +++ b/ctdb/lib/tevent/tevent_util.c @@ -107,7 +107,7 @@ void tevent_before_wait(struct event_context *ev) { if (!tevent_timeval_is_zero(&tevent_after_wait_ts)) { diff = tevent_timeval_until(&tevent_after_wait_ts, &now); if (diff.tv_sec > 3) { - tevent_debug(ev, TEVENT_DEBUG_FATAL, __location__ + tevent_debug(ev, TEVENT_DEBUG_ERROR, __location__ " Handling event took %d seconds!", (int) diff.tv_sec); } diff --git a/ctdb/libctdb/ctdb.c b/ctdb/libctdb/ctdb.c index 98c86f600d..72841dbd74 100644 --- a/ctdb/libctdb/ctdb.c +++ b/ctdb/libctdb/ctdb.c @@ -1134,3 +1134,32 @@ bool ctdb_traverse_async(struct ctdb_db *ctdb_db, return true; } + +int ctdb_num_out_queue(struct ctdb_connection *ctdb) +{ + struct ctdb_request *req; + int i; + + for (i = 0, req = ctdb->outq; req; req = req->next, i++) + ; + + return i; +} + +int ctdb_num_in_flight(struct ctdb_connection *ctdb) +{ + struct ctdb_request *req; + int i; + + for (i = 0, req = ctdb->doneq; req; req = req->next, i++) + ; + + return i; +} + +int ctdb_num_active(struct ctdb_connection *ctdb) +{ + return ctdb_num_out_queue(ctdb) + + ctdb_num_in_flight(ctdb); +} + diff --git a/ctdb/packaging/RPM/ctdb.spec.in b/ctdb/packaging/RPM/ctdb.spec.in index f618e76013..4b2d98b3b5 100644 --- a/ctdb/packaging/RPM/ctdb.spec.in +++ b/ctdb/packaging/RPM/ctdb.spec.in @@ -146,6 +146,8 @@ development libraries for ctdb %{_libdir}/libctdb.a %changelog +* Thu Sep 1 2011 : Version 1.11 + - Major updates. * Tue May 25 2010 : Version 1.10 - New version 1.10 * Tue May 25 2010 : Version 1.9 diff --git a/ctdb/server/ctdb_logging.c b/ctdb/server/ctdb_logging.c index 27b990ef67..2ef2a0b1fb 100644 --- a/ctdb/server/ctdb_logging.c +++ b/ctdb/server/ctdb_logging.c @@ -163,6 +163,7 @@ int start_syslog_daemon(struct ctdb_context *ctdb) struct ctdb_log_state { struct ctdb_context *ctdb; + const char *prefix; int fd, pfd; char buf[1024]; uint16_t buf_used; @@ -366,7 +367,11 @@ static void write_to_log(struct ctdb_log_state *log, const char *buf, unsigned int len) { if (script_log_level <= LogLevel) { - do_debug("%*.*s\n", len, len, buf); + if (log != NULL && log->prefix != NULL) { + do_debug("%s: %*.*s\n", log->prefix, len, len, buf); + } else { + do_debug("%*.*s\n", len, len, buf); + } /* log it in the eventsystem as well */ if (log->logfn) log->logfn(log->buf, len, log->logfn_private); @@ -386,7 +391,7 @@ static void ctdb_log_handler(struct event_context *ev, struct fd_event *fde, if (!(flags & EVENT_FD_READ)) { return; } - + n = read(log->pfd, &log->buf[log->buf_used], sizeof(log->buf) - log->buf_used); if (n > 0) { @@ -436,6 +441,7 @@ static int log_context_destructor(struct ctdb_log_state *log) */ struct ctdb_log_state *ctdb_fork_with_logging(TALLOC_CTX *mem_ctx, struct ctdb_context *ctdb, + const char *log_prefix, void (*logfn)(const char *, uint16_t, void *), void *logfn_private, pid_t *pid) { @@ -446,6 +452,7 @@ struct ctdb_log_state *ctdb_fork_with_logging(TALLOC_CTX *mem_ctx, log = talloc_zero(mem_ctx, struct ctdb_log_state); CTDB_NO_MEMORY_NULL(ctdb, log); log->ctdb = ctdb; + log->prefix = log_prefix; log->logfn = logfn; log->logfn_private = (void *)logfn_private; diff --git a/ctdb/server/ctdb_ltdb_server.c b/ctdb/server/ctdb_ltdb_server.c index 555d949ece..b76ae6a622 100644 --- a/ctdb/server/ctdb_ltdb_server.c +++ b/ctdb/server/ctdb_ltdb_server.c @@ -1080,6 +1080,12 @@ int32_t ctdb_control_db_attach(struct ctdb_context *ctdb, TDB_DATA indata, struct ctdb_node *node = ctdb->nodes[ctdb->pnn]; struct ctdb_client *client = NULL; + if (ctdb->tunable.allow_client_db_attach == 0) { + DEBUG(DEBUG_ERR, ("DB Attach to database %s denied by tunable " + "AllowClientDBAccess == 0\n", db_name)); + return -1; + } + /* dont allow any local clients to attach while we are in recovery mode * except for the recovery daemon. * allow all attach from the network since these are always from remote diff --git a/ctdb/server/ctdb_recoverd.c b/ctdb/server/ctdb_recoverd.c index 44a9e4bfad..631f53e89b 100644 --- a/ctdb/server/ctdb_recoverd.c +++ b/ctdb/server/ctdb_recoverd.c @@ -842,7 +842,7 @@ static void vacuum_fetch_handler(struct ctdb_context *ctdb, uint64_t srvid, } /* attach to it */ - ctdb_db = ctdb_attach(ctdb, name, persistent, 0); + ctdb_db = ctdb_attach(ctdb, CONTROL_TIMEOUT(), name, persistent, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,(__location__ " Failed to attach to database '%s'\n", name)); talloc_free(tmp_ctx); diff --git a/ctdb/server/ctdb_takeover.c b/ctdb/server/ctdb_takeover.c index 1e2e829e4d..4114b40f83 100644 --- a/ctdb/server/ctdb_takeover.c +++ b/ctdb/server/ctdb_takeover.c @@ -880,6 +880,16 @@ static int ctdb_add_public_address(struct ctdb_context *ctdb, int i; int ret; + tmp = talloc_strdup(vnn, ifaces); + for (iface = strtok(tmp, ","); iface; iface = strtok(NULL, ",")) { + if (!ctdb_sys_check_iface_exists(iface)) { + DEBUG(DEBUG_CRIT,("Interface %s does not exist. Can not add public-address : %s\n", iface, ctdb_addr_to_str(addr))); + talloc_free(tmp); + return -1; + } + } + talloc_free(tmp); + /* Verify that we dont have an entry for this ip yet */ for (vnn=ctdb->vnn;vnn;vnn=vnn->next) { if (ctdb_same_sockaddr(addr, &vnn->public_address)) { @@ -941,6 +951,51 @@ int ctdb_set_event_script_dir(struct ctdb_context *ctdb, const char *script_dir) return 0; } +static void ctdb_check_interfaces_event(struct event_context *ev, struct timed_event *te, + struct timeval t, void *private_data) +{ + struct ctdb_context *ctdb = talloc_get_type(private_data, + struct ctdb_context); + struct ctdb_vnn *vnn; + + for (vnn=ctdb->vnn;vnn;vnn=vnn->next) { + int i; + + for (i=0; vnn->ifaces[i] != NULL; i++) { + if (!ctdb_sys_check_iface_exists(vnn->ifaces[i])) { + DEBUG(DEBUG_CRIT,("Interface %s does not exist but is used by public ip %s\n", + vnn->ifaces[i], + ctdb_addr_to_str(&vnn->public_address))); + } + } + } + + event_add_timed(ctdb->ev, ctdb->check_public_ifaces_ctx, + timeval_current_ofs(30, 0), + ctdb_check_interfaces_event, ctdb); +} + + +static int ctdb_start_monitoring_interfaces(struct ctdb_context *ctdb) +{ + if (ctdb->check_public_ifaces_ctx != NULL) { + talloc_free(ctdb->check_public_ifaces_ctx); + ctdb->check_public_ifaces_ctx = NULL; + } + + ctdb->check_public_ifaces_ctx = talloc_new(ctdb); + if (ctdb->check_public_ifaces_ctx == NULL) { + ctdb_fatal(ctdb, "failed to allocate context for checking interfaces"); + } + + event_add_timed(ctdb->ev, ctdb->check_public_ifaces_ctx, + timeval_current_ofs(30, 0), + ctdb_check_interfaces_event, ctdb); + + return 0; +} + + /* setup the public address lists from a file */ @@ -1003,6 +1058,9 @@ int ctdb_set_public_addresses(struct ctdb_context *ctdb, const char *alist) } } + + ctdb_start_monitoring_interfaces(ctdb); + talloc_free(lines); return 0; } diff --git a/ctdb/server/ctdb_tunables.c b/ctdb/server/ctdb_tunables.c index 9da3cc8065..ef86051cec 100644 --- a/ctdb/server/ctdb_tunables.c +++ b/ctdb/server/ctdb_tunables.c @@ -68,7 +68,8 @@ static const struct { { "UseStatusEvents", 0, offsetof(struct ctdb_tunable, use_status_events_for_monitoring) }, { "AllowUnhealthyDBRead", 0, offsetof(struct ctdb_tunable, allow_unhealthy_db_read) }, { "StatHistoryInterval", 1, offsetof(struct ctdb_tunable, stat_history_interval) }, - { "DeferredAttachTO", 120, offsetof(struct ctdb_tunable, deferred_attach_timeout) } + { "DeferredAttachTO", 120, offsetof(struct ctdb_tunable, deferred_attach_timeout) }, + { "AllowClientDBAttach", 1, offsetof(struct ctdb_tunable, allow_client_db_attach) } }; /* diff --git a/ctdb/server/eventscript.c b/ctdb/server/eventscript.c index eb5c2da2fe..c01fbad69e 100644 --- a/ctdb/server/eventscript.c +++ b/ctdb/server/eventscript.c @@ -382,7 +382,7 @@ static int fork_child_for_script(struct ctdb_context *ctdb, return -errno; } - if (!ctdb_fork_with_logging(state, ctdb, log_event_script_output, + if (!ctdb_fork_with_logging(state, ctdb, current->name, log_event_script_output, state, &state->child)) { r = -errno; close(state->fd[0]); diff --git a/ctdb/tests/eventscripts/common.sh b/ctdb/tests/eventscripts/common.sh index ab7abb9fac..9003b396c2 100644 --- a/ctdb/tests/eventscripts/common.sh +++ b/ctdb/tests/eventscripts/common.sh @@ -1,5 +1,8 @@ # Hey Emacs, this is a -*- shell-script -*- !!! :-) +# Print a message and exit. +die () { echo "$@" >&2 ; exit 1 ; } + # Augment PATH with relevant stubs/ directories. We do this by actually # setting PATH, and also by setting $EVENTSCRIPTS_PATH and then # prepending that to $PATH in rc.local to avoid the PATH reset in @@ -30,8 +33,7 @@ if [ -d "${EVENTSCRIPTS_TESTCASE_DIR}/etc" ] ; then elif [ -d "${EVENTSCRIPTS_TESTS_DIR}/etc" ] ; then CTDB_ETCDIR="${EVENTSCRIPTS_TESTS_DIR}/etc" else - echo "Unable to set \$CTDB_ETCDIR" >&2 - exit 1 + die "Unable to set \$CTDB_ETCDIR" fi export CTDB_ETCDIR @@ -42,8 +44,7 @@ elif [ -d "${EVENTSCRIPTS_TESTCASE_DIR}/etc/ctdb" ] ; then elif [ -d "${EVENTSCRIPTS_TESTS_DIR}/etc-ctdb" ] ; then CTDB_BASE="${EVENTSCRIPTS_TESTS_DIR}/etc-ctdb" else - echo "Unable to set \$CTDB_BASE" >&2 - exit 1 + die "Unable to set \$CTDB_BASE" fi export CTDB_BASE @@ -109,7 +110,7 @@ tcp_port_down () { for _i ; do debug "Marking TCP port \"${_i}\" as not listening" - FAKE_NETSTAT_TCP_LISTEN=$(echo "$FAKE_NETSTAT_TCP_LISTEN" | sed -r -e "s@[[:space:]]*[\.0-9]+:${_i}@@g") + FAKE_TCP_LISTEN=$(echo "$FAKE_TCP_LISTEN" | sed -r -e "s@[[:space:]]*[\.0-9]+:${_i}@@g") done } @@ -177,6 +178,11 @@ ethtool_interfaces_up () done } +setup_nmap_output_filter () +{ + OUT_FILTER="-e 's@^(DEBUG: # Nmap 5.21 scan initiated) .+ (as:)@\1 DATE \2@' -e 's@^(DEBUG: # Nmap done at) .+ (--)@\1 DATE \2@'" +} + ###################################################################### # CTDB fakery @@ -232,6 +238,10 @@ setup_ctdb () export FAKE_CTDB_NODES_DOWN="$EVENTSCRIPTS_TESTS_VAR_DIR/nodes-down" mkdir -p "$FAKE_CTDB_NODES_DOWN" rm -f "$FAKE_CTDB_NODES_DOWN"/* + + export FAKE_CTDB_SCRIPTSTATUS="$EVENTSCRIPTS_TESTS_VAR_DIR/scriptstatus" + mkdir -p "$FAKE_CTDB_SCRIPTSTATUS" + rm -f "$FAKE_CTDB_SCRIPTSTATUS"/* } @@ -280,6 +290,25 @@ ctdb_get_1_public_address () echo "${_addrs%% *}" | sed -r -e 's#(.*)/(.*)@(.*)#\3 \1 \2#g' } +ctdb_not_implemented () +{ + export CTDB_NOT_IMPLEMENTED="$1" + ctdb_not_implemented="\ +DEBUG: ctdb: command \"$1\" not implemented in stub" +} + +ctdb_fake_scriptstatus () +{ + _code="$1" + _status="$2" + _err_out="$3" + + _d1=$(date '+%s.%N') + _d2=$(date '+%s.%N') + + echo "$_code $_status $_err_out" >"$FAKE_CTDB_SCRIPTSTATUS/$script" +} + ###################################################################### # Samba fakery @@ -303,8 +332,13 @@ setup_samba () export CTDB_SAMBA_SKIP_SHARE_CHECK="no" export CTDB_MANAGED_SERVICES="foo samba winbind bar" - export FAKE_NETSTAT_TCP_LISTEN="0.0.0.0:445 0.0.0.0:139" + export FAKE_TCP_LISTEN="0.0.0.0:445 0.0.0.0:139" export FAKE_WBINFO_FAIL="no" + + # Some things in 50.samba are backgrounded and waited for. If + # we don't sleep at all then timeouts can happen. This avoids + # that... :-) + export FAKE_SLEEP_FORCE=0.1 else debug "Marking Samba services as down, not listening and not managed by CTDB" # Get into known state. @@ -321,7 +355,7 @@ setup_samba () unset CTDB_MANAGES_SAMBA unset CTDB_MANAGES_WINBIND - export FAKE_NETSTAT_TCP_LISTEN="" + export FAKE_TCP_LISTEN="" export FAKE_WBINFO_FAIL="yes" fi @@ -403,9 +437,7 @@ rpc_services_up () rquotad) _t="1:2" ;; nlockmgr) _t="3:4" ;; status) _t="1:1" ;; - *) - echo "Internal error - unsupported RPC service \"${_i}\"" - exit 1 + *) die "Internal error - unsupported RPC service \"${_i}\"" ;; esac FAKE_RPCINFO_SERVICES="${FAKE_RPCINFO_SERVICES}${FAKE_RPCINFO_SERVICES:+ }${_i}:${_t}" @@ -533,8 +565,7 @@ Starting nfslock: OK" setup_vsftpd () { if [ "$1" != "down" ] ; then - echo "setup_vsftpd up not implemented!!!" - exit 1 + die "setup_vsftpd up not implemented!!!" else debug "Setting up VSFTPD environment: service down, not managed by CTDB" @@ -553,8 +584,7 @@ setup_vsftpd () setup_httpd () { if [ "$1" != "down" ] ; then - echo "setup_httpd up not implemented!!!" - exit 1 + die "setup_httpd up not implemented!!!" else debug "Setting up HTTPD environment: service down, not managed by CTDB" @@ -578,16 +608,27 @@ define_test () desc="$1" _f="$0" - _f="${_f#./}" # strip leading ./ - _f="${_f#simple/}" # strip leading simple/ - _f="${_f%%/*}" # if subdir, strip off file - _f="${_f%.sh}" # strip off .sh suffix if any + _f="${_f#./}" # strip leading ./ + _f="${_f#simple/}" # strip leading simple/ + _f="${_f#multievent/}" # strip leading multievent/ + _f="${_f%%/*}" # if subdir, strip off file + _f="${_f%.sh}" # strip off .sh suffix if any - # Remaining format should be NN.service.event.NNN: + # Remaining format should be NN.service.event.NNN or NN.service.NNN: _num="${_f##*.}" _f="${_f%.*}" - event="${_f##*.}" - script="${_f%.*}" + case "$_f" in + *.*.*) + script="${_f%.*}" + event="${_f##*.}" + ;; + *.*) + script="$_f" + unset event + ;; + *) + die "Internal error - unknown testcase filename format" + esac printf "%-14s %-10s %-4s - %s\n\n" "$script" "$event" "$_num" "$desc" } @@ -637,25 +678,24 @@ result_print () if [ -n "$_iteration" ] ; then cat <<EOF -################################################## -################################################## +================================================== Iteration $_iteration EOF fi cat <<EOF -################################################## +-------------------------------------------------- Output (Exit status: ${_rc}): -################################################## +-------------------------------------------------- $_out EOF fi if ! $_passed ; then cat <<EOF -################################################## +-------------------------------------------------- Required output (Exit status: ${required_rc}): -################################################## +-------------------------------------------------- $required_output EOF fi @@ -668,20 +708,19 @@ result_footer () if [ "$EVENTSCRIPT_TESTS_VERBOSE" = "yes" ] || ! $_passed ; then cat <<EOF -################################################## +-------------------------------------------------- CTDB_BASE="$CTDB_BASE" CTDB_ETCDIR="$CTDB_ETCDIR" ctdb client is "$(which ctdb)" -################################################## +-------------------------------------------------- EOF fi - echo - if $_passed ; then echo "PASSED" return 0 else + echo echo "FAILED" return 1 fi @@ -699,11 +738,19 @@ EOF # useful for debugging. simple_test () { - echo "Running \"${CTDB_BASE}/events.d/$script $event\"" + [ -n "$event" ] || die 'simple_test: $event not set' + + echo "Running \"$script $event${1:+ }$*\"" _out=$($EVENTSCRIPTS_TESTS_TRACE "${CTDB_BASE}/events.d/$script" "$event" "$@" 2>&1) _rc=$? - if [ "$_out" = "$required_output" -a $_rc = $required_rc ] ; then + if [ -n "$OUT_FILTER" ] ; then + _fout=$(echo "$_out" | eval sed -r $OUT_FILTER) + else + _fout="$_out" + fi + + if [ "$_fout" = "$required_output" -a $_rc = $required_rc ] ; then _passed=true else _passed=false @@ -713,6 +760,17 @@ simple_test () result_footer "$_passed" } +simple_test_event () +{ + # If something has previously failed then don't continue. + : ${_passed:=true} + $_passed || return 1 + + event="$1" ; shift + echo "##################################################" + simple_test "$@" +} + # Run an eventscript iteratively. # - 1st argument is the number of iterations. # - 2nd argument is something to eval to do setup for every iteration. @@ -733,6 +791,8 @@ simple_test () # iteration. iterate_test () { + [ -n "$event" ] || die 'simple_test: $event not set' + args="" if [ "$1" = "--" ] ; then shift @@ -747,7 +807,7 @@ iterate_test () _setup_default="$2" shift 2 - echo "Running $_repeats iterations of \"${CTDB_BASE}/events.d/$script $event\" $args" + echo "Running $_repeats iterations of \"$script $event\" $args" _result=true @@ -765,7 +825,13 @@ iterate_test () _out=$($EVENTSCRIPTS_TESTS_TRACE "${CTDB_BASE}/events.d/$script" "$event" $args 2>&1) _rc=$? - if [ "$_out" = "$required_output" -a $_rc = $required_rc ] ; then + if [ -n "$OUT_FILTER" ] ; then + _fout=$(echo "$_out" | eval sed -r $OUT_FILTER) + else + _fout="$_out" + fi + + if [ "$_fout" = "$required_output" -a $_rc = $required_rc ] ; then _passed=true else _passed=false diff --git a/ctdb/tests/eventscripts/etc/sysconfig/ctdb b/ctdb/tests/eventscripts/etc/sysconfig/ctdb new file mode 120000 index 0000000000..1c6e2b9f05 --- /dev/null +++ b/ctdb/tests/eventscripts/etc/sysconfig/ctdb @@ -0,0 +1 @@ +../../../../config/ctdb.sysconfig
\ No newline at end of file diff --git a/ctdb/tests/eventscripts/multievent/10.interface.001.sh b/ctdb/tests/eventscripts/multievent/10.interface.001.sh new file mode 100755 index 0000000000..5ef254c536 --- /dev/null +++ b/ctdb/tests/eventscripts/multievent/10.interface.001.sh @@ -0,0 +1,14 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "takeip, removeip" + +setup_ctdb + +public_address=$(ctdb_get_1_public_address) + +ok_null + +simple_test_event "takeip" $public_address +simple_test_event "releaseip" $public_address diff --git a/ctdb/tests/eventscripts/multievent/60.nfs.001.sh b/ctdb/tests/eventscripts/multievent/60.nfs.001.sh new file mode 100755 index 0000000000..a6608db83e --- /dev/null +++ b/ctdb/tests/eventscripts/multievent/60.nfs.001.sh @@ -0,0 +1,21 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "takeip, ipreallocated -> reconfigure" + +setup_nfs + +public_address=$(ctdb_get_1_public_address) + +ok_null + +simple_test_event "takeip" $public_address + +ok <<EOF +Reconfiguring service "nfs"... +Starting nfslock: OK +Starting nfs: OK +EOF + +simple_test_event "ipreallocated" diff --git a/ctdb/tests/eventscripts/multievent/60.nfs.002.sh b/ctdb/tests/eventscripts/multievent/60.nfs.002.sh new file mode 100755 index 0000000000..93f645a618 --- /dev/null +++ b/ctdb/tests/eventscripts/multievent/60.nfs.002.sh @@ -0,0 +1,25 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "takeip, monitor -> reconfigure" + +setup_nfs + +public_address=$(ctdb_get_1_public_address) + +ok_null + +simple_test_event "takeip" $public_address + +# This currently assumes that ctdb scriptstatus will always return a +# good status (when replaying). That should change and we will need +# to split this into 2 tests. +ok <<EOF +Reconfiguring service "nfs"... +Starting nfslock: OK +Starting nfs: OK +Replaying previous status for this script due to reconfigure... +EOF + +simple_test_event "monitor" diff --git a/ctdb/tests/eventscripts/multievent/60.nfs.003.sh b/ctdb/tests/eventscripts/multievent/60.nfs.003.sh new file mode 100755 index 0000000000..929eaf2ae5 --- /dev/null +++ b/ctdb/tests/eventscripts/multievent/60.nfs.003.sh @@ -0,0 +1,27 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "takeip, monitor -> reconfigure, replay error" + +setup_nfs + +public_address=$(ctdb_get_1_public_address) + +err="foo: bar error occurred" + +ok_null + +simple_test_event "takeip" $public_address + +ctdb_fake_scriptstatus 1 "ERROR" "$err" + +required_result 1 <<EOF +Reconfiguring service "nfs"... +Starting nfslock: OK +Starting nfs: OK +Replaying previous status for this script due to reconfigure... +$err +EOF + +simple_test_event "monitor" diff --git a/ctdb/tests/eventscripts/multievent/60.nfs.004.sh b/ctdb/tests/eventscripts/multievent/60.nfs.004.sh new file mode 100755 index 0000000000..addbf44c9c --- /dev/null +++ b/ctdb/tests/eventscripts/multievent/60.nfs.004.sh @@ -0,0 +1,27 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "takeip, monitor -> reconfigure, replay timedout" + +setup_nfs + +public_address=$(ctdb_get_1_public_address) + +err="waiting, waiting..." + +ok_null + +simple_test_event "takeip" $public_address + +ctdb_fake_scriptstatus -62 "TIMEDOUT" "$err" + +required_result 1 <<EOF +Reconfiguring service "nfs"... +Starting nfslock: OK +Starting nfs: OK +Replaying previous status for this script due to reconfigure... +[Replay of TIMEDOUT scriptstatus - note incorrect return code.] $err +EOF + +simple_test_event "monitor" diff --git a/ctdb/tests/eventscripts/multievent/60.nfs.005.sh b/ctdb/tests/eventscripts/multievent/60.nfs.005.sh new file mode 100755 index 0000000000..9e652013ca --- /dev/null +++ b/ctdb/tests/eventscripts/multievent/60.nfs.005.sh @@ -0,0 +1,27 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "takeip, monitor -> reconfigure, replay disabled" + +setup_nfs + +public_address=$(ctdb_get_1_public_address) + +err="" + +ok_null + +simple_test_event "takeip" $public_address + +ctdb_fake_scriptstatus -8 "DISABLED" "$err" + +ok <<EOF +Reconfiguring service "nfs"... +Starting nfslock: OK +Starting nfs: OK +Replaying previous status for this script due to reconfigure... +[Replay of DISABLED scriptstatus - note incorrect return code.] $err +EOF + +simple_test_event "monitor" diff --git a/ctdb/tests/eventscripts/multievent/60.nfs.006.sh b/ctdb/tests/eventscripts/multievent/60.nfs.006.sh new file mode 100755 index 0000000000..35f8b836e1 --- /dev/null +++ b/ctdb/tests/eventscripts/multievent/60.nfs.006.sh @@ -0,0 +1,21 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "reconfigure (synthetic), twice" +# This checks that the lock is released... + +setup_nfs + +public_address=$(ctdb_get_1_public_address) + +err="" + +ok <<EOF +Reconfiguring service "nfs"... +Starting nfslock: OK +Starting nfs: OK +EOF + +simple_test_event "reconfigure" +simple_test_event "reconfigure" diff --git a/ctdb/tests/eventscripts/run_tests.sh b/ctdb/tests/eventscripts/run_tests.sh index cb2c8f2950..ec17b5f7a7 100755 --- a/ctdb/tests/eventscripts/run_tests.sh +++ b/ctdb/tests/eventscripts/run_tests.sh @@ -15,6 +15,12 @@ for i ; do export EVENTSCRIPT_TESTS_VERBOSE="yes" shift ;; + -T) + # This will cause tests to fail but is good for debugging + # individual tests when they fail. + export EVENTSCRIPTS_TESTS_TRACE="sh -x" + shift + ;; -*) opts="$opts $i" shift diff --git a/ctdb/tests/eventscripts/simple/40.vsftpd.monitor.001.sh b/ctdb/tests/eventscripts/simple/40.vsftpd.monitor.001.sh index fe2afd1d9f..0f8b1a53c6 100755 --- a/ctdb/tests/eventscripts/simple/40.vsftpd.monitor.001.sh +++ b/ctdb/tests/eventscripts/simple/40.vsftpd.monitor.001.sh @@ -8,4 +8,4 @@ setup_vsftpd "down" ok_null -simple_test $cmd +simple_test diff --git a/ctdb/tests/eventscripts/simple/41.httpd.monitor.001.sh b/ctdb/tests/eventscripts/simple/41.httpd.monitor.001.sh index c46a45c0c7..0a9ceac428 100755 --- a/ctdb/tests/eventscripts/simple/41.httpd.monitor.001.sh +++ b/ctdb/tests/eventscripts/simple/41.httpd.monitor.001.sh @@ -8,4 +8,4 @@ setup_httpd "down" ok_null -simple_test $cmd +simple_test diff --git a/ctdb/tests/eventscripts/simple/50.samba.monitor.107.sh b/ctdb/tests/eventscripts/simple/50.samba.monitor.107.sh new file mode 100755 index 0000000000..e4151f3743 --- /dev/null +++ b/ctdb/tests/eventscripts/simple/50.samba.monitor.107.sh @@ -0,0 +1,18 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "port 139 down, default tcp checker, debug" + +# This has to go before the setup, otherwise it will write a dud file. +export CTDB_DEBUGLEVEL=4 + +setup_samba +tcp_port_down 139 + +required_result 1 <<EOF +ERROR: samba tcp port 139 is not responding +DEBUG: "ctdb checktcpport 139" was able to bind to port +EOF + +simple_test diff --git a/ctdb/tests/eventscripts/simple/50.samba.monitor.108.sh b/ctdb/tests/eventscripts/simple/50.samba.monitor.108.sh new file mode 100755 index 0000000000..d06a16f90a --- /dev/null +++ b/ctdb/tests/eventscripts/simple/50.samba.monitor.108.sh @@ -0,0 +1,15 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "port 139 down, ctdb checktcpport not implemented" + +# TODO: create nmap stub +export CTDB_NOT_IMPLEMENTED="checktcpport" + +setup_samba +tcp_port_down 139 + +required_result 1 "ERROR: samba tcp port 139 is not responding" + +simple_test diff --git a/ctdb/tests/eventscripts/simple/50.samba.monitor.109.sh b/ctdb/tests/eventscripts/simple/50.samba.monitor.109.sh new file mode 100755 index 0000000000..4bc854b623 --- /dev/null +++ b/ctdb/tests/eventscripts/simple/50.samba.monitor.109.sh @@ -0,0 +1,30 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "port 139 down, ctdb checktcpport not implemented, debug" + +ctdb_not_implemented "checktcpport" + +# This has to go before the setup, otherwise it will write a dud file. +export CTDB_DEBUGLEVEL=4 + +setup_nmap_output_filter + +setup_samba +tcp_port_down 139 + +required_result 1 <<EOF +DEBUG: ctdb_check_ports - checker ctdb not implemented +DEBUG: output from checker was: +DEBUG: ctdb checktcpport 445 (exited with 1) with output: +$ctdb_not_implemented +ERROR: samba tcp port 139 is not responding +DEBUG: nmap -n -oG - -PS 127.0.0.1 -p 445,139 shows this output: +DEBUG: # Nmap 5.21 scan initiated DATE as: nmap -n -oG - -PS 127.0.0.1 -p 445,139 +DEBUG: Host: 127.0.0.1 () Status: Up +DEBUG: Host: 127.0.0.1 () Ports: 445/open/tcp//microsoft-ds///, 139/closed/tcp//netbios-ssn/// +DEBUG: # Nmap done at DATE -- 1 IP address (1 host up) scanned in 0.04 seconds +EOF + +simple_test diff --git a/ctdb/tests/eventscripts/simple/50.samba.monitor.110.sh b/ctdb/tests/eventscripts/simple/50.samba.monitor.110.sh new file mode 100755 index 0000000000..80264cc24e --- /dev/null +++ b/ctdb/tests/eventscripts/simple/50.samba.monitor.110.sh @@ -0,0 +1,33 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "port 139 down, ctdb checktcpport/nmap not implemented, debug" + +ctdb_not_implemented "checktcpport" +export FAKE_NMAP_NOT_FOUND="yes" + +# This has to go before the setup, otherwise it will write a dud file. +export CTDB_DEBUGLEVEL=4 + +setup_nmap_output_filter + +setup_samba +tcp_port_down 139 + +required_result 1 <<EOF +DEBUG: ctdb_check_ports - checker ctdb not implemented +DEBUG: output from checker was: +DEBUG: ctdb checktcpport 445 (exited with 1) with output: +$ctdb_not_implemented +DEBUG: ctdb_check_ports - checker nmap not implemented +DEBUG: output from checker was: +DEBUG: sh: nmap: command not found +ERROR: samba tcp port 139 is not responding +DEBUG: netstat -l -t -n shows this output: +DEBUG: Active Internet connections (servers only) +DEBUG: Proto Recv-Q Send-Q Local Address Foreign Address State +DEBUG: tcp 0 0 0.0.0.0:445 0.0.0.0:* LISTEN +EOF + +simple_test diff --git a/ctdb/tests/eventscripts/simple/50.samba.monitor.111.sh b/ctdb/tests/eventscripts/simple/50.samba.monitor.111.sh new file mode 100755 index 0000000000..4651290663 --- /dev/null +++ b/ctdb/tests/eventscripts/simple/50.samba.monitor.111.sh @@ -0,0 +1,20 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "port 139 down, ctdb checktcpport/nmap/netstat not implemented" + +ctdb_not_implemented "checktcpport" +export FAKE_NMAP_NOT_FOUND="yes" +export FAKE_NETSTAT_NOT_FOUND="yes" + +setup_nmap_output_filter + +setup_samba +tcp_port_down 139 + +required_result 127 <<EOF +INTERNAL ERROR: ctdb_check_ports - no working checkers in CTDB_TCP_PORT_CHECKERS="ctdb nmap netstat" +EOF + +simple_test diff --git a/ctdb/tests/eventscripts/simple/60.nfs.monitor.161.sh b/ctdb/tests/eventscripts/simple/60.nfs.monitor.161.sh new file mode 100755 index 0000000000..4abe68eaba --- /dev/null +++ b/ctdb/tests/eventscripts/simple/60.nfs.monitor.161.sh @@ -0,0 +1,13 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "2nd share missing" + +setup_nfs + +shares_missing "ERROR: nfs directory \"%s\" not available" 2 + +required_result 1 "$MISSING_SHARES_TEXT" + +simple_test diff --git a/ctdb/tests/eventscripts/simple/60.nfs.monitor.162.sh b/ctdb/tests/eventscripts/simple/60.nfs.monitor.162.sh new file mode 100755 index 0000000000..865b1c635a --- /dev/null +++ b/ctdb/tests/eventscripts/simple/60.nfs.monitor.162.sh @@ -0,0 +1,14 @@ +#!/bin/sh + +. "${EVENTSCRIPTS_TESTS_DIR}/common.sh" + +define_test "2nd share missing, skipping share checks" + +setup_nfs +export CTDB_NFS_SKIP_SHARE_CHECK="yes" + +shares_missing "ERROR: nfs directory \"%s\" not available" 2 + +ok_null + +simple_test diff --git a/ctdb/tests/eventscripts/stubs/ctdb b/ctdb/tests/eventscripts/stubs/ctdb index ce06a35cbd..155f518ddc 100755 --- a/ctdb/tests/eventscripts/stubs/ctdb +++ b/ctdb/tests/eventscripts/stubs/ctdb @@ -2,6 +2,8 @@ prog="ctdb" +not_implemented_exit_code=1 + usage () { cat >&2 <<EOF @@ -9,26 +11,28 @@ Usage: $prog [-Y] cmd A fake CTDB stub that prints items depending on the variables FAKE_CTDB_PNN (default 0) depending on command-line options. - -Note that -Y is ignored. - EOF exit 1 } -# $POSIXLY_CORRECT means that the command passed to onnode can take -# options and getopt won't reorder things to make them options to this -# script. +not_implemented () +{ + echo "${prog}: command \"$1\" not implemented in stub" >&2 + exit $not_implemented_exit_code +} + +# Don't set $POSIXLY_CORRECT here. _temp=$(getopt -n "$prog" -o "Yvh" -l help -- "$@") || \ usage eval set -- "$_temp" verbose=false +machine_readable=false while true ; do case "$1" in - -Y) shift ;; + -Y) machine_readable=true ; shift ;; -v) verbose=true ; shift ;; --) shift ; break ;; -h|--help|*) usage ;; # * shouldn't happen, so this is reasonable. @@ -51,6 +55,13 @@ setup_pstore () mkdir -p "$pstore_dir" } +# For testing backward compatibility... +for i in $CTDB_NOT_IMPLEMENTED ; do + if [ "$i" = "$1" ] ; then + not_implemented "$i" + fi +done + case "$1" in ip) # NOTE: all nodes share the same public addresses file. @@ -159,8 +170,58 @@ case "$1" in exit 1 esac ;; + getdebug) + case "${CTDB_DEBUGLEVEL:-0}" in + -3) _t="EMERG" ;; + -2) _t="ALERT" ;; + -1) _t="CRIT" ;; + 0) _t="ERR" ;; + 1) _t="WARNING" ;; + 2) _t="NOTICE" ;; + 3) _t="INFO" ;; + 4) _t="DEBUG" ;; + *) _t="ERR" ;; + esac + + cat<<EOF +:Name:Level: +:${_t}:${CTDB_DEBUGLEVEL}: +EOF + ;; + checktcpport) + for _i in $FAKE_TCP_LISTEN ; do + if [ "$2" = "${_i##*:}" ] ; then + exit 98 + fi + done + + exit 0 + ;; + scriptstatus) + $machine_readable || not_implemented "$1, without -Y" + [ "$2" != "all" ] || not_implemented "scriptstatus all" + # For now just assume everything is good. + echo ":Type:Name:Code:Status:Start:End:Error Output...:" + for _i in "$CTDB_BASE/events.d/"*.* ; do + _d1=$(date '+%s.%N') + _b="${_i##*/}" # basename + + _f="$FAKE_CTDB_SCRIPTSTATUS/$_b" + if [ -r "$_f" ] ; then + read _code _status _err_out <"$_f" + else + _code="0" + _status="OK" + if [ ! -x "$_i" ] ; then + _status="DISABLED" + _code="-8" + fi + _err_out="" + fi + _d2=$(date '+%s.%N') + echo ":${2:-monitor}:${_b}:${_code}:${_status}:${_d1}:${_d2}:${_err_out}:" + done + ;; *) - echo "${prog}: command \"$1\" not implemented in stub" - exit 1 + not_implemented "$1" esac - diff --git a/ctdb/tests/eventscripts/stubs/exportfs b/ctdb/tests/eventscripts/stubs/exportfs index 72a11d5741..46c65225be 100755 --- a/ctdb/tests/eventscripts/stubs/exportfs +++ b/ctdb/tests/eventscripts/stubs/exportfs @@ -3,5 +3,11 @@ opts="10.0.0.0/16(rw,async,insecure,no_root_squash,no_subtree_check)" for i in $FAKE_SHARES ; do - echo "${i} ${opts}" + # Directories longer than 15 characters are printed on their own + # line. + if [ ${#i} -ge 15 ] ; then + printf '%s\n\t\t%s\n' "$i" "$opts" + else + printf '%s\t%s\n' "$i" "$opts" + fi done diff --git a/ctdb/tests/eventscripts/stubs/netstat b/ctdb/tests/eventscripts/stubs/netstat index cf0656651a..f46c936519 100755 --- a/ctdb/tests/eventscripts/stubs/netstat +++ b/ctdb/tests/eventscripts/stubs/netstat @@ -2,13 +2,20 @@ prog="netstat" +# Pretty that we're the shell and that this command could not be +# found. +if [ "$FAKE_NETSTAT_NOT_FOUND" = "yes" ] ; then + echo "sh: ${prog}: command not found" >&2 + exit 127 +fi + usage () { cat >&2 <<EOF Usage: $prog [ -t | --unix ] [ -n ] [ -a ] [ -l ] A fake netstat stub that prints items depending on the variables -FAKE_NETSTAT_TCP_ESTABLISHED, FAKE_NETSTAT_TCP_LISTEN, +FAKE_NETSTAT_TCP_ESTABLISHED, FAKE_TCP_LISTEN, FAKE_NETSTAT_UNIX_LISTEN, depending on command-line options. Note that -n is ignored. @@ -73,7 +80,7 @@ if $tcp ; then done if $all || $listen ; then - for i in $FAKE_NETSTAT_TCP_LISTEN ; do + for i in $FAKE_TCP_LISTEN ; do printf "$tcp_fmt" $i "0.0.0.0:*" "LISTEN" done fi diff --git a/ctdb/tests/eventscripts/stubs/nmap b/ctdb/tests/eventscripts/stubs/nmap new file mode 100755 index 0000000000..f01fe32d9c --- /dev/null +++ b/ctdb/tests/eventscripts/stubs/nmap @@ -0,0 +1,75 @@ +#!/bin/bash + +prog="nmap" + +# Pretty that we're the shell and that this command could not be +# found. +if [ "$FAKE_NMAP_NOT_FOUND" = "yes" ] ; then + echo "sh: ${prog}: command not found" >&2 + exit 127 +fi + +usage () +{ + cat >&2 <<EOF +Usage: $prog -n -oG - -PS 127.0.0.1 -p <port>[,<port> ...] + +A fake nmap stub that prints items depending on the variable +FAKE_TCP_LISTEN and the ports specified. + +Note that all options apart from -p are ignored. + +EOF + exit 1 +} + +ports="" + +parse_options () +{ + _temp=$(getopt -n "$prog" -a -o "np:" -l help -l PS: -l oG: -- "$@") + + [ $? != 0 ] && usage + + eval set -- "$_temp" + + while true ; do + case "$1" in + -n) shift ;; + --oG|--PS) shift 2 ;; + -p) ports="${ports}${ports:+ }${2//,/ }" ; shift 2 ;; + --) shift ; break ;; + -h|--help|*) usage ;; # * shouldn't happen, so this is reasonable. + esac + done + + [ $# -gt 0 ] && usage + + [ -n "$ports" ] || usage +} + +# For printing out... +args="$*" + +parse_options "$@" + +port_states="" + +for p in $ports ; do + pn=$(getent services "$p" | sed -e 's@[[:space:]].*@@') + for i in $FAKE_TCP_LISTEN ; do + lp="${i##*:}" + if [ "$p" = "$lp" ] ; then + port_states="${port_states}${port_states:+, }${p}/open/tcp//${pn}///" + continue 2 + fi + done + port_states="${port_states}${port_states:+, }${p}/closed/tcp//${pn}///" +done + +cat <<EOF +# Nmap 5.21 scan initiated $(date) as: nmap $args +Host: 127.0.0.1 () Status: Up +Host: 127.0.0.1 () Ports: $port_states +# Nmap done at $(date) -- 1 IP address (1 host up) scanned in 0.04 seconds +EOF diff --git a/ctdb/tests/eventscripts/stubs/sleep b/ctdb/tests/eventscripts/stubs/sleep new file mode 100755 index 0000000000..e4542444de --- /dev/null +++ b/ctdb/tests/eventscripts/stubs/sleep @@ -0,0 +1,9 @@ +#!/bin/sh + +if [ "$FAKE_SLEEP_REALLY" = "yes" ] ; then + /bin/sleep "$@" +elif [ -n "$FAKE_SLEEP_FORCE" ] ; then + /bin/sleep "$FAKE_SLEEP_FORCE" +else + : +fi diff --git a/ctdb/tests/src/ctdb_bench.c b/ctdb/tests/src/ctdb_bench.c index 6d812237ec..8e0e0d7956 100644 --- a/ctdb/tests/src/ctdb_bench.c +++ b/ctdb/tests/src/ctdb_bench.c @@ -222,10 +222,11 @@ int main(int argc, const char *argv[]) ev = event_context_init(NULL); /* initialise ctdb */ - ctdb = ctdb_cmdline_client(ev); + ctdb = ctdb_cmdline_client(ev, timeval_current_ofs(3, 0)); /* attach to a specific database */ - ctdb_db = ctdb_attach(ctdb, "test.tdb", false, 0); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), "test.tdb", + false, 0); if (!ctdb_db) { printf("ctdb_attach failed - %s\n", ctdb_errstr(ctdb)); exit(1); diff --git a/ctdb/tests/src/ctdb_fetch.c b/ctdb/tests/src/ctdb_fetch.c index a8f38eeea3..f0311c8948 100644 --- a/ctdb/tests/src/ctdb_fetch.c +++ b/ctdb/tests/src/ctdb_fetch.c @@ -207,7 +207,7 @@ int main(int argc, const char *argv[]) ev = event_context_init(NULL); - ctdb = ctdb_cmdline_client(ev); + ctdb = ctdb_cmdline_client(ev, timeval_current_ofs(3, 0)); if (ctdb == NULL) { printf("failed to connect to ctdb daemon.\n"); @@ -218,7 +218,8 @@ int main(int argc, const char *argv[]) &cluster_ready); /* attach to a specific database */ - ctdb_db = ctdb_attach(ctdb, "test.tdb", false, 0); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), "test.tdb", + false, 0); if (!ctdb_db) { printf("ctdb_attach failed - %s\n", ctdb_errstr(ctdb)); exit(1); diff --git a/ctdb/tests/src/ctdb_fetch_one.c b/ctdb/tests/src/ctdb_fetch_one.c index 7348aa6986..042e605f96 100644 --- a/ctdb/tests/src/ctdb_fetch_one.c +++ b/ctdb/tests/src/ctdb_fetch_one.c @@ -114,7 +114,7 @@ int main(int argc, const char *argv[]) ev = event_context_init(NULL); - ctdb = ctdb_cmdline_client(ev); + ctdb = ctdb_cmdline_client(ev, timeval_current_ofs(3, 0)); if (ctdb == NULL) { printf("failed to connect to ctdb deamon.\n"); @@ -122,7 +122,8 @@ int main(int argc, const char *argv[]) } /* attach to a specific database */ - ctdb_db = ctdb_attach(ctdb, "test.tdb", false, 0); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), "test.tdb", + false, 0); if (!ctdb_db) { printf("ctdb_attach failed - %s\n", ctdb_errstr(ctdb)); exit(1); diff --git a/ctdb/tests/src/ctdb_persistent.c b/ctdb/tests/src/ctdb_persistent.c index 6074b6a1e3..98c100f4ed 100644 --- a/ctdb/tests/src/ctdb_persistent.c +++ b/ctdb/tests/src/ctdb_persistent.c @@ -220,7 +220,7 @@ int main(int argc, const char *argv[]) ev = event_context_init(NULL); - ctdb = ctdb_cmdline_client(ev); + ctdb = ctdb_cmdline_client(ev, timeval_current_ofs(3, 0)); if (ctdb == NULL) { printf("Could not attach to daemon\n"); return 1; @@ -228,9 +228,11 @@ int main(int argc, const char *argv[]) /* attach to a specific database */ if (unsafe_writes == 1) { - ctdb_db = ctdb_attach(ctdb, "persistent.tdb", true, TDB_NOSYNC); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), + "persistent.tdb", true, TDB_NOSYNC); } else { - ctdb_db = ctdb_attach(ctdb, "persistent.tdb", true, 0); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), + "persistent.tdb", true, 0); } if (!ctdb_db) { diff --git a/ctdb/tests/src/ctdb_randrec.c b/ctdb/tests/src/ctdb_randrec.c index 5b7eef83d2..dbaa4ba0fc 100644 --- a/ctdb/tests/src/ctdb_randrec.c +++ b/ctdb/tests/src/ctdb_randrec.c @@ -181,7 +181,7 @@ int main(int argc, const char *argv[]) ev = event_context_init(NULL); - ctdb = ctdb_cmdline_client(ev); + ctdb = ctdb_cmdline_client(ev, timeval_current_ofs(3, 0)); if (ctdb == NULL) { printf("failed to connect to daemon\n"); @@ -189,7 +189,8 @@ int main(int argc, const char *argv[]) } /* attach to a specific database */ - ctdb_db = ctdb_attach(ctdb, "test.tdb", false, 0); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), "test.tdb", + false, 0); if (!ctdb_db) { printf("ctdb_attach failed - %s\n", ctdb_errstr(ctdb)); exit(1); diff --git a/ctdb/tests/src/ctdb_store.c b/ctdb/tests/src/ctdb_store.c index 82669b7b6d..c657f7850e 100644 --- a/ctdb/tests/src/ctdb_store.c +++ b/ctdb/tests/src/ctdb_store.c @@ -136,7 +136,7 @@ int main(int argc, const char *argv[]) ev = event_context_init(NULL); - ctdb = ctdb_cmdline_client(ev); + ctdb = ctdb_cmdline_client(ev, timeval_current_ofs(3, 0)); if (ctdb == NULL) { printf("failed to connect to ctdb daemon.\n"); @@ -144,7 +144,7 @@ int main(int argc, const char *argv[]) } /* attach to a specific database */ - ctdb_db = ctdb_attach(ctdb, "test.tdb", false, 0); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), "test.tdb", false, 0); if (!ctdb_db) { printf("ctdb_attach failed - %s\n", ctdb_errstr(ctdb)); exit(1); diff --git a/ctdb/tests/src/ctdb_transaction.c b/ctdb/tests/src/ctdb_transaction.c index 39e40dd16b..6508874717 100644 --- a/ctdb/tests/src/ctdb_transaction.c +++ b/ctdb/tests/src/ctdb_transaction.c @@ -252,7 +252,7 @@ int main(int argc, const char *argv[]) ev = event_context_init(NULL); - ctdb = ctdb_cmdline_client(ev); + ctdb = ctdb_cmdline_client(ev, timeval_current_ofs(3, 0)); if (ctdb == NULL) { DEBUG(DEBUG_ERR, ("Could not attach to daemon\n")); return 1; @@ -260,9 +260,11 @@ int main(int argc, const char *argv[]) /* attach to a specific database */ if (unsafe_writes == 1) { - ctdb_db = ctdb_attach(ctdb, "transaction.tdb", true, TDB_NOSYNC); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), + "transaction.tdb", true, TDB_NOSYNC); } else { - ctdb_db = ctdb_attach(ctdb, "transaction.tdb", true, 0); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), + "transaction.tdb", true, 0); } if (!ctdb_db) { diff --git a/ctdb/tests/src/ctdb_traverse.c b/ctdb/tests/src/ctdb_traverse.c index c4510e43c9..218e7f777d 100644 --- a/ctdb/tests/src/ctdb_traverse.c +++ b/ctdb/tests/src/ctdb_traverse.c @@ -89,10 +89,10 @@ int main(int argc, const char *argv[]) ev = event_context_init(NULL); - ctdb = ctdb_cmdline_client(ev); + ctdb = ctdb_cmdline_client(ev, timeval_current_ofs(3, 0)); /* attach to a specific database */ - ctdb_db = ctdb_attach(ctdb, dbname, false, 0); + ctdb_db = ctdb_attach(ctdb, timeval_current_ofs(2, 0), dbname, false, 0); if (!ctdb_db) { printf("ctdb_attach failed - %s\n", ctdb_errstr(ctdb)); exit(1); diff --git a/ctdb/tools/ctdb.c b/ctdb/tools/ctdb.c index 331392c2e4..26c0bfb769 100644 --- a/ctdb/tools/ctdb.c +++ b/ctdb/tools/ctdb.c @@ -624,9 +624,10 @@ static int control_status(struct ctdb_context *ctdb, int argc, const char **argv return ret; } - if(options.machinereadable){ - printf(":Node:IP:Disconnected:Banned:Disabled:Unhealthy:Stopped:Inactive:PartiallyOnline:\n"); - for(i=0;i<nodemap->num;i++){ + if (options.machinereadable) { + printf(":Node:IP:Disconnected:Banned:Disabled:Unhealthy:Stopped" + ":Inactive:PartiallyOnline:ThisNode:\n"); + for (i=0;i<nodemap->num;i++) { int partially_online = 0; int j; @@ -650,7 +651,7 @@ static int control_status(struct ctdb_context *ctdb, int argc, const char **argv talloc_free(ifaces); } } - printf(":%d:%s:%d:%d:%d:%d:%d:%d:%d:\n", nodemap->nodes[i].pnn, + printf(":%d:%s:%d:%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), @@ -658,7 +659,8 @@ static int control_status(struct ctdb_context *ctdb, int argc, const char **argv !!(nodemap->nodes[i].flags&NODE_FLAGS_UNHEALTHY), !!(nodemap->nodes[i].flags&NODE_FLAGS_STOPPED), !!(nodemap->nodes[i].flags&NODE_FLAGS_INACTIVE), - partially_online); + partially_online, + (nodemap->nodes[i].pnn == mypnn)); } return 0; } @@ -737,17 +739,15 @@ static int control_status(struct ctdb_context *ctdb, int argc, const char **argv printf("hash:%d lmaster:%d\n", i, vnnmap->map[i]); } - ret = ctdb_ctrl_getrecmode(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmode); - if (ret != 0) { + if (!ctdb_getrecmode(ctdb_connection, options.pnn, &recmode)) { DEBUG(DEBUG_ERR, ("Unable to get recmode from node %u\n", options.pnn)); - return ret; + return -1; } printf("Recovery mode:%s (%d)\n",recmode==CTDB_RECOVERY_NORMAL?"NORMAL":"RECOVERY",recmode); - ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster); - if (ret != 0) { + if (!ctdb_getrecmaster(ctdb_connection, options.pnn, &recmaster)) { DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn)); - return ret; + return -1; } printf("Recovery master:%d\n",recmaster); @@ -961,7 +961,7 @@ static int control_one_scriptstatus(struct ctdb_context *ctdb, break; } if (options.machinereadable) { - printf("%s:%s:%i:%s:%lu.%06lu:%lu.%06lu:%s:\n", + printf(":%s:%s:%i:%s:%lu.%06lu:%lu.%06lu:%s:\n", ctdb_eventscript_call_names[type], script_status->scripts[i].name, script_status->scripts[i].status, @@ -1095,13 +1095,11 @@ static int control_disablescript(struct ctdb_context *ctdb, int argc, const char */ static int control_recmaster(struct ctdb_context *ctdb, int argc, const char **argv) { - int ret; uint32_t recmaster; - ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster); - if (ret != 0) { + if (!ctdb_getrecmaster(ctdb_connection, options.pnn, &recmaster)) { DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn)); - return ret; + return -1; } printf("%d\n",recmaster); @@ -1596,10 +1594,9 @@ again: } - ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster); - if (ret != 0) { + if (!ctdb_getrecmaster(ctdb_connection, options.pnn, &recmaster)) { DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn)); - return ret; + return -1; } /* verify the node exists */ @@ -2543,15 +2540,13 @@ static uint32_t get_generation(struct ctdb_context *ctdb) } /* get the recmaster */ - ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), CTDB_CURRENT_NODE, &recmaster); - if (ret != 0) { + if (!ctdb_getrecmaster(ctdb_connection, CTDB_CURRENT_NODE, &recmaster)) { DEBUG(DEBUG_ERR, ("Unable to get recmaster from node %u\n", options.pnn)); exit(10); } /* get recovery mode */ - ret = ctdb_ctrl_getrecmode(ctdb, ctdb, TIMELIMIT(), recmaster, &recmode); - if (ret != 0) { + if (!ctdb_getrecmode(ctdb_connection, recmaster, &recmode)) { DEBUG(DEBUG_ERR, ("Unable to get recmode from node %u\n", options.pnn)); exit(10); } @@ -2990,7 +2985,7 @@ static int control_catdb(struct ctdb_context *ctdb, int argc, const char **argv) return -1; } - ctdb_db = ctdb_attach(ctdb, db_name, persistent, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name)); @@ -3048,7 +3043,7 @@ static int control_cattdb(struct ctdb_context *ctdb, int argc, const char **argv return -1; } - ctdb_db = ctdb_attach(ctdb, db_name, false, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, false, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name)); @@ -3092,7 +3087,7 @@ static int control_readkey(struct ctdb_context *ctdb, int argc, const char **arg return -1; } - ctdb_db = ctdb_attach(ctdb, db_name, persistent, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name)); @@ -3141,7 +3136,7 @@ static int control_writekey(struct ctdb_context *ctdb, int argc, const char **ar return -1; } - ctdb_db = ctdb_attach(ctdb, db_name, persistent, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name)); @@ -3205,7 +3200,7 @@ static int control_pfetch(struct ctdb_context *ctdb, int argc, const char **argv return -1; } - ctdb_db = ctdb_attach(ctdb, db_name, persistent, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name)); @@ -3368,8 +3363,7 @@ static int control_pstore(struct ctdb_context *ctdb, int argc, const char **argv db_name = argv[0]; - ctdb_db = ctdb_attach(ctdb, db_name, true, 0); - + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, true, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name)); talloc_free(tmp_ctx); @@ -3657,7 +3651,6 @@ static int control_getdbstatus(struct ctdb_context *ctdb, int argc, const char * static int control_isnotrecmaster(struct ctdb_context *ctdb, int argc, const char **argv) { uint32_t mypnn, recmaster; - int ret; mypnn = ctdb_ctrl_getpnn(ctdb, TIMELIMIT(), options.pnn); if (mypnn == -1) { @@ -3665,8 +3658,7 @@ static int control_isnotrecmaster(struct ctdb_context *ctdb, int argc, const cha return 1; } - ret = ctdb_ctrl_getrecmaster(ctdb, ctdb, TIMELIMIT(), options.pnn, &recmaster); - if (ret != 0) { + if (!ctdb_getrecmaster(ctdb_connection, options.pnn, &recmaster)) { printf("Failed to get the recmaster\n"); return 1; } @@ -4023,7 +4015,7 @@ static int control_attach(struct ctdb_context *ctdb, int argc, const char **argv persistent = true; } - ctdb_db = ctdb_attach(ctdb, db_name, persistent, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), db_name, persistent, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", db_name)); return -1; @@ -4241,7 +4233,7 @@ static int control_backupdb(struct ctdb_context *ctdb, int argc, const char **ar DEBUG(DEBUG_ERR,("database '%s' is unhealthy: %s\n", argv[0], reason)); - DEBUG(DEBUG_ERR,("disallow backup : tunnable AllowUnhealthyDBRead = %u\n", + DEBUG(DEBUG_ERR,("disallow backup : tunable AllowUnhealthyDBRead = %u\n", allow_unhealthy)); talloc_free(tmp_ctx); return -1; @@ -4254,7 +4246,7 @@ static int control_backupdb(struct ctdb_context *ctdb, int argc, const char **ar allow_unhealthy)); } - ctdb_db = ctdb_attach(ctdb, argv[0], dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), argv[0], dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", argv[0])); talloc_free(tmp_ctx); @@ -4401,7 +4393,7 @@ static int control_restoredb(struct ctdb_context *ctdb, int argc, const char **a dbname, tbuf); - ctdb_db = ctdb_attach(ctdb, dbname, dbhdr.persistent, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), dbname, dbhdr.persistent, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,("Unable to attach to database '%s'\n", dbname)); talloc_free(tmp_ctx); @@ -4653,7 +4645,7 @@ static int control_wipedb(struct ctdb_context *ctdb, int argc, return -1; } - ctdb_db = ctdb_attach(ctdb, argv[0], dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), argv[0], dbmap->dbs[i].flags & CTDB_DB_FLAGS_PERSISTENT, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR, ("Unable to attach to database '%s'\n", argv[0])); @@ -5232,7 +5224,7 @@ int main(int argc, const char *argv[]) } /* initialise ctdb */ - ctdb = ctdb_cmdline_client(ev); + ctdb = ctdb_cmdline_client(ev, TIMELIMIT()); if (ctdb_commands[i].without_daemon == false) { const char *socket_name; diff --git a/ctdb/tools/ctdb_vacuum.c b/ctdb/tools/ctdb_vacuum.c index cbe3c5c8bb..a9dc28a9a6 100644 --- a/ctdb/tools/ctdb_vacuum.c +++ b/ctdb/tools/ctdb_vacuum.c @@ -260,7 +260,7 @@ static int ctdb_vacuum_db(struct ctdb_context *ctdb, uint32_t db_id, struct ctdb return -1; } - ctdb_db = ctdb_attach(ctdb, name, persistent, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), name, persistent, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,(__location__ " Failed to attach to database '%s'\n", name)); talloc_free(vdata); @@ -581,7 +581,7 @@ static int ctdb_repack_db(struct ctdb_context *ctdb, uint32_t db_id, return -1; } - ctdb_db = ctdb_attach(ctdb, name, persistent, 0); + ctdb_db = ctdb_attach(ctdb, TIMELIMIT(), name, persistent, 0); if (ctdb_db == NULL) { DEBUG(DEBUG_ERR,(__location__ " Failed to attach to database '%s'\n", name)); return -1; diff --git a/ctdb/utils/pmda/Install b/ctdb/utils/pmda/Install new file mode 100644 index 0000000000..a56a63506a --- /dev/null +++ b/ctdb/utils/pmda/Install @@ -0,0 +1,36 @@ +#! /bin/sh +# +# Copyright (c) 1997 Silicon Graphics, Inc. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or modify it +# under the terms of the GNU General Public License as published by the +# Free Software Foundation; either version 2 of the License, or (at your +# option) any later version. +# +# This program is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +# or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +# for more details. +# +# You should have received a copy of the GNU General Public License along +# with this program; if not, write to the Free Software Foundation, Inc., +# 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA +# +# Install the ctdb PMDA and/or PMNS +# + +. $PCP_DIR/etc/pcp.env +. $PCP_SHARE_DIR/lib/pmdaproc.sh + +iam=ctdb +pmda_interface=2 + +# runs as daemon and only supports pipe IPC +daemon_opt=true +dso_opt=false +pipe_opt=true +socket_opt=false + +pmdaSetup +pmdaInstall +exit 0 diff --git a/ctdb/utils/pmda/README b/ctdb/utils/pmda/README new file mode 100644 index 0000000000..f8dbbbccac --- /dev/null +++ b/ctdb/utils/pmda/README @@ -0,0 +1,84 @@ +CTDB PMDA +=========== + +This PMDA extracts metrics from the locally running ctdbd daemon for +export to PMCD. + +Note: + This PMDA may be remade from source and hence requires IDO (or + more specifically a C compiler) to be installed. + + Uses of make(1) may fail (without removing or clobbering files) + if the C compiler cannot be found. This is most likely to + happen when running the PMDA ./Install script. + + The only remedial action is to install the C compiler, or + hand-craft changes to the Makefile. + +Metrics +======= + +The file ./help contains descriptions for all of the metrics exported +by this PMDA. + +Once the PMDA has been installed, the following command will list all +the available metrics and their explanatory "help" text: + + $ pminfo -fT ctdb + +Installation +============ + + + # cd $PCP_PMDAS_DIR/ctdb + + + Check that there is no clash in the Performance Metrics Domain + defined in ./domain.h and the other PMDAs currently in use (see + $PCP_PMCDCONF_PATH). If there is, edit ./domain.h to choose another + domain number. + + + Then simply use + + # ./Install + + and choose both the "collector" and "monitor" installation + configuration options. + + You will be prompted to choose either a daemon implementation + or a DSO implementation of the PMDA, and in the case of the daemon + variant to select an IPC method -- everything else is automated + +De-installation +=============== + + + Simply use + + # cd $PCP_PMDAS_DIR/ctdb + # ./Remove + +Troubleshooting +=============== + + + After installing or restarting the agent, the PMCD log file + ($PCP_LOG_DIR/pmcd/pmcd.log) and the PMDA log file + ($PCP_LOG_DIR/pmcd/pmda_ctdb.log) should be checked for any warnings + or errors. + + +Adding a New Metric +=================== + +This section walks through the development task of adding a new metric to the +CTDB PMDA. + + + Define the metric in the pmns file with a unique metric id. See the pmns(4) + man page for details. + + + Add a description of the metric to the help file. + + + Taking note of the previously assigned metric id, add a new entry to the + metrictab structure in pmda_ctdb.c. See the pmdaInit(3) man page for + details. + + + Ensure the counter is already a member of the ctdb_statistics structure. + Finally, add code to pmda_ctdb_fetch_cb() to handle fetch requests for the + newly defined metric. diff --git a/ctdb/utils/pmda/Remove b/ctdb/utils/pmda/Remove new file mode 100644 index 0000000000..7d1c509e50 --- /dev/null +++ b/ctdb/utils/pmda/Remove @@ -0,0 +1,29 @@ +#! /bin/sh +# +# Copyright (c) 1997 Silicon Graphics, Inc. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or modify it +# under the terms of the GNU General Public License as published by the +# Free Software Foundation; either version 2 of the License, or (at your +# option) any later version. +# +# This program is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +# or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +# for more details. +# +# You should have received a copy of the GNU General Public License along +# with this program; if not, write to the Free Software Foundation, Inc., +# 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA +# +# Remove the ctdb PMDA +# + +. $PCP_DIR/etc/pcp.env +. $PCP_SHARE_DIR/lib/pmdaproc.sh + +iam=ctdb + +pmdaSetup +pmdaRemove +exit 0 diff --git a/ctdb/utils/pmda/domain.h b/ctdb/utils/pmda/domain.h new file mode 100644 index 0000000000..d9b316f66c --- /dev/null +++ b/ctdb/utils/pmda/domain.h @@ -0,0 +1,19 @@ +/* domain.h + * + * Copyright (c) 2004-2009 Silicon Graphics, Inc. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +#define CTDB 155 diff --git a/ctdb/utils/pmda/help b/ctdb/utils/pmda/help new file mode 100644 index 0000000000..0e9984eab3 --- /dev/null +++ b/ctdb/utils/pmda/help @@ -0,0 +1,106 @@ +# +# Copyright (c) 2000-2004 Silicon Graphics, Inc. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or modify it +# under the terms of the GNU General Public License as published by the +# Free Software Foundation; either version 2 of the License, or (at your +# option) any later version. +# +# This program is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +# or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +# for more details. +# +# You should have received a copy of the GNU General Public License along +# with this program; if not, write to the Free Software Foundation, Inc., +# 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA +# +# ctdb PMDA help file in the ASCII format +# +# lines beginning with a # are ignored +# lines beginning @ introduce a new entry of the form +# @ metric_name oneline-text +# help test goes +# here over multiple lines +# ... +# +# the metric_name is decoded against the default PMNS -- as a special case, +# a name of the form NNN.MM (for numeric NNN and MM) is interpreted as an +# instance domain identification, and the text describes the instance domain +# +# blank lines before the @ line are ignored +# + +@ ctdb.num_clients number of clients connected to ctdbd + +@ ctdb.frozen whether any databases are frozen + +@ ctdb.recovering whether recovery is active + +@ ctdb.client_packets_sent number of packets sent to all clients + +@ ctdb.client_packets_recv number of packets received from all clients + +@ ctdb.node_packets_sent number of packets sent to other nodes + +@ ctdb.node_packets_recv number of packets received from other nodes + +@ ctdb.keepalive_packets_sent number of keepalive packets sent to other nodes + +@ ctdb.keepalive_packets_recv number of keepalive packets received from other nodes + +@ ctdb.node.req_call number of node CTDB_REQ_CALL packets handled + +@ ctdb.node.reply_call number of node CTDB_REPLY_CALL packets handled + +@ ctdb.node.req_dmaster number of node CTDB_REQ_DMASTER packets handled + +@ ctdb.node.reply_dmaster number of node CTDB_REPLY_DMASTER packets handled + +@ ctdb.node.reply_error number of node CTDB_REPLY_ERROR packets handled + +@ ctdb.node.req_message number of node CTDB_REQ_MESSAGE packets handled + +@ ctdb.node.req_control number of node CTDB_REQ_CONTROL packets handled + +@ ctdb.node.reply_control number of node CTDB_REPLY_CONTROL packets handled + +@ ctdb.client.req_call number of client CTDB_REQ_CALL packets handled + +@ ctdb.client.req_message number of client CTDB_REQ_MESSAGE packets handled + +@ ctdb.client.req_control number of client CTDB_REQ_CONTROL packets handled + +@ ctdb.timeouts.call (counter not implemented) number of call timeouts + +@ ctdb.timeouts.control number of node control message request timeouts awaiting reply + +@ ctdb.timeouts.traverse number of database traversal timeouts + +@ ctdb.total_calls total number of client ctdb request calls received + +@ ctdb.pending_calls total number of client ctdb request calls in progress + +@ ctdb.lockwait_calls number of tdb chainlock lockwait calls + +@ ctdb.pending_lockwait_calls number of lockwait calls waiting for a lock + +@ ctdb.childwrite_calls number of childwrite calls + +@ ctdb.pending_childwrite_calls number of childwrite calls in progress + +@ ctdb.memory_used total size of the ctdbd null talloc pool + +@ ctdb.max_hop_count maximum hops performed by a CTDB_REQ_CALL packet + +@ ctdb.max_reclock_ctdbd maximum recovery lock latency during setrecmode + +@ ctdb.max_reclock_recd maximum recovery lock latency as reported by the recovery process + +@ ctdb.max_call_latency maximum time spent handling a client request call + +@ ctdb.max_lockwait_latency maximum time spent waiting for a tdb chainlock + +@ ctdb.max_childwrite_latency maximum time spent performing a childwrite + +@ ctdb.num_recoveries number of recoveries finished diff --git a/ctdb/utils/pmda/pmda_ctdb.c b/ctdb/utils/pmda/pmda_ctdb.c new file mode 100644 index 0000000000..927fea5805 --- /dev/null +++ b/ctdb/utils/pmda/pmda_ctdb.c @@ -0,0 +1,596 @@ +/* + * CTDB Performance Metrics Domain Agent (PMDA) for Performance Co-Pilot (PCP) + * + * Copyright (c) 1995,2004 Silicon Graphics, Inc. All Rights Reserved. + * Copyright (c) 2011 David Disseldorp + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2 of the License, or (at your + * option) any later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY + * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License + * for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include <pcp/pmapi.h> +#include <pcp/impl.h> +#include <pcp/pmda.h> +#include "../../include/includes.h" +#include "../../lib/tevent/tevent.h" +#include "../../include/ctdb.h" +#include "../../include/ctdb_private.h" +#include "../../include/ctdb_protocol.h" +#include "domain.h" + +/* + * CTDB PMDA + * + * This PMDA connects to the locally running ctdbd daemon and pulls + * statistics for export via PCP. The ctdbd Unix domain socket path can be + * specified with the CTDB_SOCKET environment variable, otherwise the default + * path is used. + */ + +/* + * All metrics supported in this PMDA - one table entry for each. + * The 4th field specifies the serial number of the instance domain + * for the metric, and must be either PM_INDOM_NULL (denoting a + * metric that only ever has a single value), or the serial number + * of one of the instance domains declared in the instance domain table + * (i.e. in indomtab, above). + */ +static pmdaMetric metrictab[] = { + /* num_clients */ + { NULL, { PMDA_PMID(0,0), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,0,0,0,0,0) }, }, + /* frozen */ + { NULL, { PMDA_PMID(1,2), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,0,0,0,0,0) }, }, + /* recovering */ + { NULL, { PMDA_PMID(3,3), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,0,0,0,0,0) }, }, + /* client_packets_sent */ + { NULL, { PMDA_PMID(4,4), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* client_packets_recv */ + { NULL, { PMDA_PMID(5,5), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* node_packets_sent */ + { NULL, { PMDA_PMID(6,6), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* node_packets_recv */ + { NULL, { PMDA_PMID(7,7), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* keepalive_packets_sent */ + { NULL, { PMDA_PMID(8,8), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* keepalive_packets_recv */ + { NULL, { PMDA_PMID(9,9), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* req_call */ + { NULL, { PMDA_PMID(10,10), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* reply_call */ + { NULL, { PMDA_PMID(10,11), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* req_dmaster */ + { NULL, { PMDA_PMID(10,12), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* reply_dmaster */ + { NULL, { PMDA_PMID(10,13), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* reply_error */ + { NULL, { PMDA_PMID(10,14), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* req_message */ + { NULL, { PMDA_PMID(10,15), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* req_control */ + { NULL, { PMDA_PMID(10,16), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* reply_control */ + { NULL, { PMDA_PMID(10,17), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* req_call */ + { NULL, { PMDA_PMID(11,18), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* req_message */ + { NULL, { PMDA_PMID(11,19), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* req_control */ + { NULL, { PMDA_PMID(11,20), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* call */ + { NULL, { PMDA_PMID(12,21), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,0) }, }, + /* control */ + { NULL, { PMDA_PMID(12,22), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,0) }, }, + /* traverse */ + { NULL, { PMDA_PMID(12,23), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,0) }, }, + /* total_calls */ + { NULL, { PMDA_PMID(13,24), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* pending_calls */ + { NULL, { PMDA_PMID(14,25), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,0,0,0,0,0) }, }, + /* lockwait_calls */ + { NULL, { PMDA_PMID(15,27), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* pending_lockwait_calls */ + { NULL, { PMDA_PMID(16,27), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,0,0,0,0,0) }, }, + /* childwrite_calls */ + { NULL, { PMDA_PMID(17,28), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_COUNTER, + PMDA_PMUNITS(0,0,1,0,0,PM_COUNT_ONE) }, }, + /* pending_childwrite_calls */ + { NULL, { PMDA_PMID(18,29), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,0,0,0,0,0) }, }, + /* memory_used */ + { NULL, { PMDA_PMID(19,30), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(1,0,0,PM_SPACE_BYTE,0,0) }, }, + /* max_hop_count */ + { NULL, { PMDA_PMID(20,31), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,0,0,0,0,0) }, }, + /* max_reclock_ctdbd */ + { NULL, { PMDA_PMID(21,32), PM_TYPE_DOUBLE, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,1,0,0,PM_TIME_SEC,0) }, }, + /* max_reclock_recd */ + { NULL, { PMDA_PMID(22,33), PM_TYPE_DOUBLE, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,1,0,0,PM_TIME_SEC,0) }, }, + /* max_call_latency */ + { NULL, { PMDA_PMID(23,34), PM_TYPE_DOUBLE, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,1,0,0,PM_TIME_SEC,0) }, }, + /* max_lockwait_latency */ + { NULL, { PMDA_PMID(24,35), PM_TYPE_DOUBLE, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,1,0,0,PM_TIME_SEC,0) }, }, + /* max_childwrite_latency */ + { NULL, { PMDA_PMID(25,36), PM_TYPE_DOUBLE, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,1,0,0,PM_TIME_SEC,0) }, }, + /* num_recoveries */ + { NULL, { PMDA_PMID(26,37), PM_TYPE_U32, PM_INDOM_NULL, PM_SEM_INSTANT, + PMDA_PMUNITS(0,0,0,0,0,0) }, }, +}; + +static struct event_context *ev; +static struct ctdb_context *ctdb; +static struct ctdb_statistics *stats; + +static void +pmda_ctdb_q_read_cb(uint8_t *data, size_t cnt, void *args) +{ + if (cnt == 0) { + fprintf(stderr, "ctdbd unreachable\n"); + /* cleanup on request timeout */ + return; + } + + ctdb_client_read_cb(data, cnt, args); +} + + +static int +pmda_ctdb_daemon_connect(void) +{ + const char *socket_name; + int ret; + struct sockaddr_un addr; + + ev = event_context_init(NULL); + if (ev == NULL) { + fprintf(stderr, "Failed to init event ctx\n"); + return -1; + } + + ctdb = ctdb_init(ev); + if (ctdb == NULL) { + fprintf(stderr, "Failed to init ctdb\n"); + goto err_ev; + } + + socket_name = getenv("CTDB_SOCKET"); + if (socket_name == NULL) { + socket_name = CTDB_PATH; + } + + ret = ctdb_set_socketname(ctdb, socket_name); + if (ret == -1) { + fprintf(stderr, "ctdb_set_socketname failed - %s\n", + ctdb_errstr(ctdb)); + goto err_ctdb; + } + + /* + * ctdb_socket_connect() sets a default queue callback handler that + * calls exit() if ctdbd is unavailable on recv, use our own wrapper to + * work around this + */ + + memset(&addr, 0, sizeof(addr)); + addr.sun_family = AF_UNIX; + strncpy(addr.sun_path, ctdb->daemon.name, sizeof(addr.sun_path)); + + ctdb->daemon.sd = socket(AF_UNIX, SOCK_STREAM, 0); + if (ctdb->daemon.sd == -1) { + fprintf(stderr, "Failed to open client socket\n"); + goto err_ctdb; + } + + set_nonblocking(ctdb->daemon.sd); + set_close_on_exec(ctdb->daemon.sd); + + if (connect(ctdb->daemon.sd, (struct sockaddr *)&addr, sizeof(addr)) == -1) { + fprintf(stderr, "Failed to connect to ctdb daemon via %s\n", + ctdb->daemon.name); + goto err_sd; + } + + ctdb->daemon.queue = ctdb_queue_setup(ctdb, ctdb, ctdb->daemon.sd, + CTDB_DS_ALIGNMENT, + pmda_ctdb_q_read_cb, ctdb, + "to-ctdbd"); + if (ctdb->daemon.queue == NULL) { + fprintf(stderr, "Failed to setup queue\n"); + goto err_sd; + } + + ctdb->pnn = ctdb_ctrl_getpnn(ctdb, timeval_current_ofs(3, 0), + CTDB_CURRENT_NODE); + if (ctdb->pnn == (uint32_t)-1) { + fprintf(stderr, "Failed to get ctdb pnn\n"); + goto err_sd; + } + + return 0; +err_sd: + close(ctdb->daemon.sd); +err_ctdb: + talloc_free(ctdb); +err_ev: + talloc_free(ev); + ctdb = NULL; + return -1; +} + +static void +pmda_ctdb_daemon_disconnect(void) +{ + if (ctdb->methods) { + ctdb->methods->shutdown(ctdb); + } + + if (ctdb->daemon.sd != -1) { + close(ctdb->daemon.sd); + } + + talloc_free(ctdb); + talloc_free(ev); + ctdb = NULL; +} + +static int +fill_node(unsigned int item, pmAtomValue *atom) +{ + switch (item) { + case 10: + atom->ul = stats->node.req_call; + break; + case 11: + atom->ul = stats->node.reply_call; + break; + case 12: + atom->ul = stats->node.req_dmaster; + break; + case 13: + atom->ul = stats->node.reply_dmaster; + break; + case 14: + atom->ul = stats->node.reply_error; + break; + case 15: + atom->ul = stats->node.req_message; + break; + case 16: + atom->ul = stats->node.req_control; + break; + case 17: + atom->ul = stats->node.reply_control; + break; + default: + return PM_ERR_PMID; + } + + return 0; +} + +static int +fill_client(unsigned int item, pmAtomValue *atom) +{ + switch (item) { + case 18: + atom->ul = stats->client.req_call; + break; + case 19: + atom->ul = stats->client.req_message; + break; + case 20: + atom->ul = stats->client.req_control; + break; + default: + return PM_ERR_PMID; + } + + return 0; +} + +static int +fill_timeout(unsigned int item, pmAtomValue *atom) +{ + switch (item) { + case 21: + atom->ul = stats->timeouts.call; + break; + case 22: + atom->ul = stats->timeouts.control; + break; + case 23: + atom->ul = stats->timeouts.traverse; + break; + default: + return PM_ERR_PMID; + } + + return 0; +} + +/* + * callback provided to pmdaFetch + */ +static int +pmda_ctdb_fetch_cb(pmdaMetric *mdesc, unsigned int inst, pmAtomValue *atom) +{ + int ret; + __pmID_int *id = (__pmID_int *)&(mdesc->m_desc.pmid); + + if (inst != PM_IN_NULL) { + return PM_ERR_INST; + } + + if (stats == NULL) { + fprintf(stderr, "stats not available\n"); + ret = PM_ERR_VALUE; + goto err_out; + } + + + switch (id->cluster) { + case 0: + atom->ul = stats->num_clients; + break; + case 1: + atom->ul = stats->frozen; + break; + case 3: + atom->ul = stats->recovering; + break; + case 4: + atom->ul = stats->client_packets_sent; + break; + case 5: + atom->ul = stats->client_packets_recv; + break; + case 6: + atom->ul = stats->node_packets_sent; + break; + case 7: + atom->ul = stats->node_packets_recv; + break; + case 8: + atom->ul = stats->keepalive_packets_sent; + break; + case 9: + atom->ul = stats->keepalive_packets_recv; + break; + case 10: + ret = fill_node(id->item, atom); + if (ret) { + goto err_out; + } + break; + case 11: + ret = fill_client(id->item, atom); + if (ret) { + goto err_out; + } + break; + case 12: + ret = fill_timeout(id->item, atom); + if (ret) { + goto err_out; + } + break; + case 13: + atom->ul = stats->total_calls; + break; + case 14: + atom->ul = stats->pending_calls; + break; + case 15: + atom->ul = stats->lockwait_calls; + break; + case 16: + atom->ul = stats->pending_lockwait_calls; + break; + case 17: + atom->ul = stats->childwrite_calls; + break; + case 18: + atom->ul = stats->pending_childwrite_calls; + break; + case 19: + atom->ul = stats->memory_used; + break; + case 20: + atom->ul = stats->max_hop_count; + break; + case 21: + atom->d = stats->reclock.ctdbd.max; + break; + case 22: + atom->d = stats->reclock.recd.max; + break; + case 23: + atom->d = stats->call_latency.max; + break; + case 24: + atom->d = stats->lockwait_latency.max; + break; + case 25: + atom->d = stats->childwrite_latency.max; + break; + case 26: + atom->d = stats->num_recoveries; + break; + default: + return PM_ERR_PMID; + } + + ret = 0; +err_out: + return ret; +} + +/* + * This routine is called once for each pmFetch(3) operation, so is a + * good place to do once-per-fetch functions, such as value caching or + * instance domain evaluation. + */ +static int +pmda_ctdb_fetch(int numpmid, pmID pmidlist[], pmResult **resp, pmdaExt *pmda) +{ + int ret; + TDB_DATA data; + int32_t res; + struct timeval ctdb_timeout; + + if (ctdb == NULL) { + fprintf(stderr, "attempting reconnect to ctdbd\n"); + ret = pmda_ctdb_daemon_connect(); + if (ret < 0) { + fprintf(stderr, "reconnect failed\n"); + return PM_ERR_VALUE; + } + } + + ctdb_timeout = timeval_current_ofs(1, 0); + ret = ctdb_control(ctdb, ctdb->pnn, 0, + CTDB_CONTROL_STATISTICS, 0, tdb_null, + ctdb, &data, &res, &ctdb_timeout, NULL); + + if (ret != 0 || res != 0) { + fprintf(stderr, "ctdb control for statistics failed, reconnecting\n"); + pmda_ctdb_daemon_disconnect(); + ret = PM_ERR_VALUE; + goto err_out; + } + + stats = (struct ctdb_statistics *)data.dptr; + + if (data.dsize != sizeof(struct ctdb_statistics)) { + fprintf(stderr, "incorrect statistics size %zu - not %zu\n", + data.dsize, sizeof(struct ctdb_statistics)); + ret = PM_ERR_VALUE; + goto err_stats; + } + + ret = pmdaFetch(numpmid, pmidlist, resp, pmda); + +err_stats: + talloc_free(stats); +err_out: + return ret; +} + +/* + * Initialise the agent + */ +void +pmda_ctdb_init(pmdaInterface *dp) +{ + if (dp->status != 0) { + return; + } + + dp->version.two.fetch = pmda_ctdb_fetch; + pmdaSetFetchCallBack(dp, pmda_ctdb_fetch_cb); + + pmdaInit(dp, NULL, 0, metrictab, + (sizeof(metrictab) / sizeof(metrictab[0]))); +} + +static char * +helpfile(void) +{ + static char buf[MAXPATHLEN]; + + if (!buf[0]) { + snprintf(buf, sizeof(buf), "%s/ctdb/help", + pmGetConfig("PCP_PMDAS_DIR")); + } + return buf; +} + +static void +usage(void) +{ + fprintf(stderr, "Usage: %s [options]\n\n", pmProgname); + fputs("Options:\n" + " -d domain use domain (numeric) for metrics domain of PMDA\n" + " -l logfile write log into logfile rather than using default log name\n" + "\nExactly one of the following options may appear:\n" + " -i port expect PMCD to connect on given inet port (number or name)\n" + " -p expect PMCD to supply stdin/stdout (pipe)\n" + " -u socket expect PMCD to connect on given unix domain socket\n", + stderr); + exit(1); +} + +/* + * Set up the agent if running as a daemon. + */ +int +main(int argc, char **argv) +{ + int err = 0; + char log_file[] = "pmda_ctdb.log"; + pmdaInterface dispatch; + + __pmSetProgname(argv[0]); + + pmdaDaemon(&dispatch, PMDA_INTERFACE_2, pmProgname, CTDB, + log_file, helpfile()); + + if (pmdaGetOpt(argc, argv, "d:i:l:pu:?", &dispatch, &err) != EOF) { + err++; + } + + if (err) { + usage(); + } + + pmdaOpenLog(&dispatch); + pmda_ctdb_init(&dispatch); + pmdaConnect(&dispatch); + pmdaMain(&dispatch); + + exit(0); +} + diff --git a/ctdb/utils/pmda/pmns b/ctdb/utils/pmda/pmns new file mode 100644 index 0000000000..38509defb9 --- /dev/null +++ b/ctdb/utils/pmda/pmns @@ -0,0 +1,73 @@ +/* + * Metrics for CTDB PMDA + * + * Copyright (c) 2000-2004 Silicon Graphics, Inc. All Rights Reserved. + * Copyright (c) 2011 David Disseldorp + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2 of the License, or (at your + * option) any later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY + * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License + * for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +ctdb { + num_clients CTDB:0:0 + frozen CTDB:1:2 + recovering CTDB:3:3 + client_packets_sent CTDB:4:4 + client_packets_recv CTDB:5:5 + node_packets_sent CTDB:6:6 + node_packets_recv CTDB:7:7 + keepalive_packets_sent CTDB:8:8 + keepalive_packets_recv CTDB:9:9 + node + client + timeouts + total_calls CTDB:13:24 + pending_calls CTDB:14:25 + lockwait_calls CTDB:15:27 + pending_lockwait_calls CTDB:16:27 + childwrite_calls CTDB:17:28 + pending_childwrite_calls CTDB:18:29 + memory_used CTDB:19:30 + max_hop_count CTDB:20:31 + max_reclock_ctdbd CTDB:21:32 + max_reclock_recd CTDB:22:33 + max_call_latency CTDB:23:34 + max_lockwait_latency CTDB:24:35 + max_childwrite_latency CTDB:25:36 + num_recoveries CTDB:26:37 +} + +ctdb.node { + req_call CTDB:10:10 + reply_call CTDB:10:11 + req_dmaster CTDB:10:12 + reply_dmaster CTDB:10:13 + reply_error CTDB:10:14 + req_message CTDB:10:15 + req_control CTDB:10:16 + reply_control CTDB:10:17 +} + +ctdb.client { + req_call CTDB:11:18 + req_message CTDB:11:19 + req_control CTDB:11:20 +} + +ctdb.timeouts { + call CTDB:12:21 + control CTDB:12:22 + traverse CTDB:12:23 +} + diff --git a/ctdb/utils/pmda/root b/ctdb/utils/pmda/root new file mode 100644 index 0000000000..ff036ed7fb --- /dev/null +++ b/ctdb/utils/pmda/root @@ -0,0 +1,10 @@ +/* + * fake "root" for validating the local PMNS subtree + */ + +#include <stdpmid> + +root { ctdb } + +#include "pmns" + |