ctdb/server/ctdbd.c

   1 /*
   2    standalone ctdb daemon
   3
   4    Copyright (C) Andrew Tridgell  2006
   5
   6    This program is free software; you can redistribute it and/or modify
   7    it under the terms of the GNU General Public License as published by
   8    the Free Software Foundation; either version 3 of the License, or
   9    (at your option) any later version.
  10
  11    This program is distributed in the hope that it will be useful,
  12    but WITHOUT ANY WARRANTY; without even the implied warranty of
  13    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14    GNU General Public License for more details.
  15
  16    You should have received a copy of the GNU General Public License
  17    along with this program; if not, see <http://www.gnu.org/licenses/>.
  18 */
  19
  20 #include "replace.h"
  21 #include "system/filesys.h"
  22 #include "system/time.h"
  23 #include "system/wait.h"
  24 #include "system/network.h"
  25
  26 #include <popt.h>
  27 #include <talloc.h>
  28 /* Allow use of deprecated function tevent_loop_allow_nesting() */
  29 #define TEVENT_DEPRECATED
  30 #include <tevent.h>
  31
  32 #include "lib/util/debug.h"
  33 #include "lib/util/samba_util.h"
  34
  35 #include "ctdb_private.h"
  36
  37 #include "common/reqid.h"
  38 #include "common/system.h"
  39 #include "common/cmdline.h"
  40 #include "common/common.h"
  41 #include "common/logging.h"
  42
  43 static struct {
  44         const char *nlist;
  45         const char *transport;
  46         const char *myaddress;
  47         const char *public_address_list;
  48         const char *event_script_dir;
  49         const char *notification_script;
  50         const char *logging;
  51         const char *recovery_lock_file;
  52         const char *db_dir;
  53         const char *db_dir_persistent;
  54         const char *db_dir_state;
  55         const char *public_interface;
  56         const char *single_public_ip;
  57         int         valgrinding;
  58         int         nosetsched;
  59         int         start_as_disabled;
  60         int         start_as_stopped;
  61         int         no_lmaster;
  62         int         no_recmaster;
  63         int         lvs;
  64         int         script_log_level;
  65         int         no_publicipcheck;
  66         int         max_persistent_check_errors;
  67 } options = {
  68         .nlist = NULL,
  69         .public_address_list = NULL,
  70         .transport = "tcp",
  71         .event_script_dir = NULL,
  72         .logging = "file:" LOGDIR "/log.ctdb",
  73         .db_dir = CTDB_VARDIR,
  74         .db_dir_persistent = CTDB_VARDIR "/persistent",
  75         .db_dir_state = CTDB_VARDIR "/state",
  76         .script_log_level = DEBUG_ERR,
  77 };
  78
  79 int script_log_level;
  80 bool fast_start;
  81
  82 /*
  83   called by the transport layer when a packet comes in
  84 */
  85 static void ctdb_recv_pkt(struct ctdb_context *ctdb, uint8_t *data, uint32_t length)
  86 {
  87         struct ctdb_req_header *hdr = (struct ctdb_req_header *)data;
  88
  89         CTDB_INCREMENT_STAT(ctdb, node_packets_recv);
  90
  91         /* up the counter for this source node, so we know its alive */
  92         if (ctdb_validate_pnn(ctdb, hdr->srcnode)) {
  93                 /* as a special case, redirected calls don't increment the rx_cnt */
  94                 if (hdr->operation != CTDB_REQ_CALL ||
  95                     ((struct ctdb_req_call_old *)hdr)->hopcount == 0) {
  96                         ctdb->nodes[hdr->srcnode]->rx_cnt++;
  97                 }
  98         }
  99
 100         ctdb_input_pkt(ctdb, hdr);
 101 }
 102
 103 static const struct ctdb_upcalls ctdb_upcalls = {
 104         .recv_pkt       = ctdb_recv_pkt,
 105         .node_dead      = ctdb_node_dead,
 106         .node_connected = ctdb_node_connected
 107 };
 108
 109
 110
 111 /*
 112   main program
 113 */
 114 int main(int argc, const char *argv[])
 115 {
 116         struct ctdb_context *ctdb;
 117         int interactive = 0;
 118
 119         struct poptOption popt_options[] = {
 120                 POPT_AUTOHELP
 121                 POPT_CTDB_CMDLINE
 122                 { "interactive", 'i', POPT_ARG_NONE, &interactive, 0, "don't fork", NULL },
 123                 { "public-addresses", 0, POPT_ARG_STRING, &options.public_address_list, 0, "public address list file", "filename" },
 124                 { "public-interface", 0, POPT_ARG_STRING, &options.public_interface, 0, "public interface", "interface"},
 125                 { "single-public-ip", 0, POPT_ARG_STRING, &options.single_public_ip, 0, "single public ip", "ip-address"},
 126                 { "event-script-dir", 0, POPT_ARG_STRING, &options.event_script_dir, 0, "event script directory", "dirname" },
 127                 { "logging", 0, POPT_ARG_STRING, &options.logging, 0, "logging method to be used", NULL },
 128                 { "nlist", 0, POPT_ARG_STRING, &options.nlist, 0, "node list file", "filename" },
 129                 { "notification-script", 0, POPT_ARG_STRING, &options.notification_script, 0, "notification script", "filename" },
 130                 { "listen", 0, POPT_ARG_STRING, &options.myaddress, 0, "address to listen on", "address" },
 131                 { "transport", 0, POPT_ARG_STRING, &options.transport, 0, "protocol transport", NULL },
 132                 { "dbdir", 0, POPT_ARG_STRING, &options.db_dir, 0, "directory for the tdb files", NULL },
 133                 { "dbdir-persistent", 0, POPT_ARG_STRING, &options.db_dir_persistent, 0, "directory for persistent tdb files", NULL },
 134                 { "dbdir-state", 0, POPT_ARG_STRING, &options.db_dir_state, 0, "directory for internal state tdb files", NULL },
 135                 { "reclock", 0, POPT_ARG_STRING, &options.recovery_lock_file, 0, "location of recovery lock file", "filename" },
 136                 { "pidfile", 0, POPT_ARG_STRING, &ctdbd_pidfile, 0, "location of PID file", "filename" },
 137                 { "valgrinding", 0, POPT_ARG_NONE, &options.valgrinding, 0, "disable setscheduler SCHED_FIFO call, use mmap for tdbs", NULL },
 138                 { "nosetsched", 0, POPT_ARG_NONE, &options.nosetsched, 0, "disable setscheduler SCHED_FIFO call, use mmap for tdbs", NULL },
 139                 { "start-as-disabled", 0, POPT_ARG_NONE, &options.start_as_disabled, 0, "Node starts in disabled state", NULL },
 140                 { "start-as-stopped", 0, POPT_ARG_NONE, &options.start_as_stopped, 0, "Node starts in stopped state", NULL },
 141                 { "no-lmaster", 0, POPT_ARG_NONE, &options.no_lmaster, 0, "disable lmaster role on this node", NULL },
 142                 { "no-recmaster", 0, POPT_ARG_NONE, &options.no_recmaster, 0, "disable recmaster role on this node", NULL },
 143                 { "lvs", 0, POPT_ARG_NONE, &options.lvs, 0, "lvs is enabled on this node", NULL },
 144                 { "script-log-level", 0, POPT_ARG_INT, &options.script_log_level, 0, "log level of event script output", NULL },
 145                 { "nopublicipcheck", 0, POPT_ARG_NONE, &options.no_publicipcheck, 0, "don't check we have/don't have the correct public ip addresses", NULL },
 146                 { "max-persistent-check-errors", 0, POPT_ARG_INT,
 147                   &options.max_persistent_check_errors, 0,
 148                   "max allowed persistent check errors (default 0)", NULL },
 149                 { "sloppy-start", 0, POPT_ARG_NONE, &fast_start, 0, "Do not perform full recovery on start", NULL },
 150                 POPT_TABLEEND
 151         };
 152         int opt, ret;
 153         const char **extra_argv;
 154         int extra_argc = 0;
 155         poptContext pc;
 156         struct tevent_context *ev;
 157
 158         pc = poptGetContext(argv[0], argc, argv, popt_options, POPT_CONTEXT_KEEP_FIRST);
 159
 160         while ((opt = poptGetNextOpt(pc)) != -1) {
 161                 switch (opt) {
 162                 default:
 163                         fprintf(stderr, "Invalid option %s: %s\n",
 164                                 poptBadOption(pc, 0), poptStrerror(opt));
 165                         exit(1);
 166                 }
 167         }
 168
 169         /* setup the remaining options for the main program to use */
 170         extra_argv = poptGetArgs(pc);
 171         if (extra_argv) {
 172                 extra_argv++;
 173                 while (extra_argv[extra_argc]) extra_argc++;
 174         }
 175
 176         talloc_enable_null_tracking();
 177
 178         fault_setup();
 179
 180         ev = tevent_context_init(NULL);
 181         if (ev == NULL) {
 182                 DEBUG(DEBUG_ALERT,("tevent_context_init() failed\n"));
 183                 exit(1);
 184         }
 185         tevent_loop_allow_nesting(ev);
 186
 187         ctdb = ctdb_cmdline_init(ev);
 188
 189         ctdb->start_as_disabled = options.start_as_disabled;
 190         ctdb->start_as_stopped  = options.start_as_stopped;
 191
 192         script_log_level = options.script_log_level;
 193
 194         if (!ctdb_logging_init(ctdb, options.logging)) {
 195                 exit(1);
 196         }
 197
 198         DEBUG(DEBUG_NOTICE,("CTDB starting on node\n"));
 199
 200         gettimeofday(&ctdb->ctdbd_start_time, NULL);
 201         gettimeofday(&ctdb->last_recovery_started, NULL);
 202         gettimeofday(&ctdb->last_recovery_finished, NULL);
 203         ctdb->recovery_mode    = CTDB_RECOVERY_NORMAL;
 204         ctdb->recovery_master  = (uint32_t)-1;
 205         ctdb->upcalls          = &ctdb_upcalls;
 206         ctdb->recovery_lock_fd = -1;
 207
 208         ret = reqid_init(ctdb, 0, &ctdb->idr);;
 209         if (ret != 0) {
 210                 DEBUG(DEBUG_ALERT, ("reqid_init failed (%s)\n", strerror(ret)));
 211                 exit(1);
 212         }
 213
 214         ctdb_tunables_set_defaults(ctdb);
 215
 216         ret = ctdb_set_recovery_lock_file(ctdb, options.recovery_lock_file);
 217         if (ret == -1) {
 218                 DEBUG(DEBUG_ALERT,("ctdb_set_recovery_lock_file failed - %s\n", ctdb_errstr(ctdb)));
 219                 exit(1);
 220         }
 221
 222         ret = ctdb_set_transport(ctdb, options.transport);
 223         if (ret == -1) {
 224                 DEBUG(DEBUG_ALERT,("ctdb_set_transport failed - %s\n", ctdb_errstr(ctdb)));
 225                 exit(1);
 226         }
 227
 228         /* tell ctdb what address to listen on */
 229         if (options.myaddress) {
 230                 ret = ctdb_set_address(ctdb, options.myaddress);
 231                 if (ret == -1) {
 232                         DEBUG(DEBUG_ALERT,("ctdb_set_address failed - %s\n", ctdb_errstr(ctdb)));
 233                         exit(1);
 234                 }
 235         }
 236
 237         /* set ctdbd capabilities */
 238         ctdb->capabilities = CTDB_CAP_DEFAULT;
 239         if (options.no_lmaster != 0) {
 240                 ctdb->capabilities &= ~CTDB_CAP_LMASTER;
 241         }
 242         if (options.no_recmaster != 0) {
 243                 ctdb->capabilities &= ~CTDB_CAP_RECMASTER;
 244         }
 245         if (options.lvs != 0) {
 246                 ctdb->capabilities |= CTDB_CAP_LVS;
 247         }
 248
 249         /* Initialise this node's PNN to the unknown value.  This will
 250          * be set to the correct value by either ctdb_add_node() as
 251          * part of loading the nodes file or by
 252          * ctdb_tcp_listen_automatic() when the transport is
 253          * initialised.  At some point we should de-optimise this and
 254          * pull it out into ctdb_start_daemon() so it is done clearly
 255          * and only in one place.
 256          */
 257         ctdb->pnn = -1;
 258
 259         /* Default value for CTDB_BASE - don't override */
 260         setenv("CTDB_BASE", CTDB_ETCDIR, 0);
 261
 262         /* tell ctdb what nodes are available */
 263         if (options.nlist != NULL) {
 264                 ctdb->nodes_file = options.nlist;
 265         } else {
 266                 ctdb->nodes_file =
 267                         talloc_asprintf(ctdb, "%s/nodes", getenv("CTDB_BASE"));
 268                 if (ctdb->nodes_file == NULL) {
 269                         DEBUG(DEBUG_ALERT,(__location__ " Out of memory\n"));
 270                         exit(1);
 271                 }
 272         }
 273         ctdb_load_nodes_file(ctdb);
 274
 275         ctdb->db_directory = options.db_dir;
 276         mkdir_p_or_die(ctdb->db_directory, 0700);
 277
 278         ctdb->db_directory_persistent = options.db_dir_persistent;
 279         mkdir_p_or_die(ctdb->db_directory_persistent, 0700);
 280
 281         ctdb->db_directory_state = options.db_dir_state;
 282         mkdir_p_or_die(ctdb->db_directory_state, 0700);
 283
 284         if (options.public_interface) {
 285                 ctdb->default_public_interface = talloc_strdup(ctdb, options.public_interface);
 286                 CTDB_NO_MEMORY(ctdb, ctdb->default_public_interface);
 287         }
 288
 289         if (options.single_public_ip) {
 290                 if (options.public_interface == NULL) {
 291                         DEBUG(DEBUG_ALERT,("--single_public_ip used but --public_interface is not specified. You must specify the public interface when using single public ip. Exiting\n"));
 292                         exit(10);
 293                 }
 294
 295                 ret = ctdb_set_single_public_ip(ctdb, options.public_interface,
 296                                                 options.single_public_ip);
 297                 if (ret != 0) {
 298                         DEBUG(DEBUG_ALERT,("Invalid --single-public-ip argument : %s . This is not a valid ip address. Exiting.\n", options.single_public_ip));
 299                         exit(10);
 300                 }
 301         }
 302
 303         if (options.event_script_dir != NULL) {
 304                 ctdb->event_script_dir = options.event_script_dir;
 305         } else {
 306                 ctdb->event_script_dir = talloc_asprintf(ctdb, "%s/events.d",
 307                                                          getenv("CTDB_BASE"));
 308                 if (ctdb->event_script_dir == NULL) {
 309                         DEBUG(DEBUG_ALERT,(__location__ " Out of memory\n"));
 310                         exit(1);
 311                 }
 312         }
 313
 314         if (options.notification_script != NULL) {
 315                 ret = ctdb_set_notification_script(ctdb, options.notification_script);
 316                 if (ret == -1) {
 317                         DEBUG(DEBUG_ALERT,("Unable to setup notification script\n"));
 318                         exit(1);
 319                 }
 320         }
 321
 322         ctdb->valgrinding = options.valgrinding;
 323         if (options.valgrinding || options.nosetsched) {
 324                 ctdb->do_setsched = 0;
 325         } else {
 326                 ctdb->do_setsched = 1;
 327         }
 328
 329         ctdb->public_addresses_file = options.public_address_list;
 330         ctdb->do_checkpublicip = !options.no_publicipcheck;
 331
 332         if (options.max_persistent_check_errors < 0) {
 333                 ctdb->max_persistent_check_errors = 0xFFFFFFFFFFFFFFFFLL;
 334         } else {
 335                 ctdb->max_persistent_check_errors = (uint64_t)options.max_persistent_check_errors;
 336         }
 337
 338         /* start the protocol running (as a child) */
 339         return ctdb_start_daemon(ctdb, interactive?false:true);
 340 }