2 * Copyright (C) 2013 - Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License, version 2 only,
6 * as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11 * GNU General Public License for more details.
13 * You should have received a copy of the GNU General Public License along
14 * with this program; if not, write to the Free Software Foundation, Inc.,
15 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
29 #include <sys/resource.h>
31 #include <sys/socket.h>
33 #include <sys/types.h>
34 #include <urcu/list.h>
39 #include <urcu/compiler.h>
42 #include <common/defaults.h>
43 #include <common/common.h>
44 #include <common/consumer/consumer.h>
45 #include <common/consumer/consumer-timer.h>
46 #include <common/compat/poll.h>
47 #include <common/sessiond-comm/sessiond-comm.h>
48 #include <common/utils.h>
49 #include <common/compat/getenv.h>
50 #include <common/fd-tracker/utils.h>
52 #include "lttng-relayd.h"
53 #include "health-relayd.h"
55 /* Global health check unix path */
57 char health_unix_sock_path
[PATH_MAX
];
59 int health_quit_pipe
[2] = { -1, -1 };
62 * Check if the thread quit pipe was triggered.
64 * Return 1 if it was triggered else 0;
67 int check_health_quit_pipe(int fd
, uint32_t events
)
69 if (fd
== health_quit_pipe
[0] && (events
& LPOLLIN
)) {
77 * Send data on a unix socket using the liblttsessiondcomm API.
79 * Return lttcomm error code.
81 static int send_unix_sock(int sock
, void *buf
, size_t len
)
83 /* Check valid length */
88 return lttcomm_send_unix_sock(sock
, buf
, len
);
91 static int create_lttng_rundir_with_perm(const char *rundir
)
95 DBG3("Creating LTTng run directory: %s", rundir
);
97 ret
= mkdir(rundir
, S_IRWXU
);
99 if (errno
!= EEXIST
) {
100 ERR("Unable to create %s", rundir
);
105 } else if (ret
== 0) {
106 int is_root
= !getuid();
111 ret
= utils_get_group_id(tracing_group_name
, true, &gid
);
113 /* Default to root group. */
117 ret
= chown(rundir
, 0, gid
);
119 ERR("Unable to set group on %s", rundir
);
126 S_IRUSR
| S_IWUSR
| S_IXUSR
| S_IRGRP
| S_IXGRP
| S_IROTH
| S_IXOTH
);
128 ERR("Unable to set permissions on %s", health_unix_sock_path
);
141 int parse_health_env(void)
143 const char *health_path
;
145 health_path
= lttng_secure_getenv(LTTNG_RELAYD_HEALTH_ENV
);
147 strncpy(health_unix_sock_path
, health_path
,
149 health_unix_sock_path
[PATH_MAX
- 1] = '\0';
156 int setup_health_path(void)
158 int is_root
, ret
= 0;
159 const char *home_path
= NULL
;
160 char *rundir
= NULL
, *relayd_path
= NULL
;
162 ret
= parse_health_env();
170 rundir
= strdup(DEFAULT_LTTNG_RUNDIR
);
177 * Create rundir from home path. This will create something like
180 home_path
= utils_get_home_dir();
182 if (home_path
== NULL
) {
183 /* TODO: Add --socket PATH option */
184 ERR("Can't get HOME directory for sockets creation.");
189 ret
= asprintf(&rundir
, DEFAULT_LTTNG_HOME_RUNDIR
, home_path
);
196 ret
= asprintf(&relayd_path
, DEFAULT_RELAYD_PATH
, rundir
);
202 ret
= create_lttng_rundir_with_perm(rundir
);
207 ret
= create_lttng_rundir_with_perm(relayd_path
);
213 if (strlen(health_unix_sock_path
) != 0) {
216 snprintf(health_unix_sock_path
, sizeof(health_unix_sock_path
),
217 DEFAULT_GLOBAL_RELAY_HEALTH_UNIX_SOCK
,
220 /* Set health check Unix path */
221 if (strlen(health_unix_sock_path
) != 0) {
225 snprintf(health_unix_sock_path
, sizeof(health_unix_sock_path
),
226 DEFAULT_HOME_RELAY_HEALTH_UNIX_SOCK
,
227 home_path
, (int) getpid());
237 int accept_unix_socket(void *data
, int *out_fd
)
240 int accepting_sock
= *((int *) data
);
242 ret
= lttcomm_accept_unix_sock(accepting_sock
);
254 * Thread managing health check socket.
256 void *thread_manage_health(void *data
)
258 int sock
= -1, new_sock
= -1, ret
, i
, pollfd
, err
= -1;
259 uint32_t revents
, nb_fd
;
260 struct lttng_poll_event events
;
261 struct health_comm_msg msg
;
262 struct health_comm_reply reply
;
265 DBG("[thread] Manage health check started");
269 rcu_register_thread();
271 /* We might hit an error path before this is created. */
272 lttng_poll_init(&events
);
274 /* Create unix socket */
275 sock
= lttcomm_create_unix_sock(health_unix_sock_path
);
277 ERR("Unable to create health check Unix socket");
284 /* lttng health client socket path permissions */
287 ret
= utils_get_group_id(tracing_group_name
, true, &gid
);
289 /* Default to root group. */
293 ret
= chown(health_unix_sock_path
, 0, gid
);
295 ERR("Unable to set group on %s", health_unix_sock_path
);
301 ret
= chmod(health_unix_sock_path
,
302 S_IRUSR
| S_IWUSR
| S_IRGRP
| S_IWGRP
);
304 ERR("Unable to set permissions on %s", health_unix_sock_path
);
312 * Set the CLOEXEC flag. Return code is useless because either way, the
315 (void) utils_set_fd_cloexec(sock
);
317 ret
= lttcomm_listen_unix_sock(sock
);
322 /* Size is set to 1 for the consumer_channel pipe */
323 ret
= lttng_poll_create(&events
, 2, LTTNG_CLOEXEC
);
325 ERR("Poll set creation failed");
329 ret
= lttng_poll_add(&events
, health_quit_pipe
[0], LPOLLIN
);
334 /* Add the application registration socket */
335 ret
= lttng_poll_add(&events
, sock
, LPOLLIN
| LPOLLPRI
);
340 lttng_relay_notify_ready();
343 char *accepted_socket_name
;
345 DBG("Health check ready");
347 /* Inifinite blocking call, waiting for transmission */
349 ret
= lttng_poll_wait(&events
, -1);
352 * Restart interrupted system call.
354 if (errno
== EINTR
) {
362 for (i
= 0; i
< nb_fd
; i
++) {
363 /* Fetch once the poll data */
364 revents
= LTTNG_POLL_GETEV(&events
, i
);
365 pollfd
= LTTNG_POLL_GETFD(&events
, i
);
367 /* Thread quit pipe has been closed. Killing thread. */
368 ret
= check_health_quit_pipe(pollfd
, revents
);
374 /* Event on the registration socket */
375 if (pollfd
== sock
) {
376 if (revents
& LPOLLIN
) {
378 } else if (revents
& (LPOLLERR
| LPOLLHUP
| LPOLLRDHUP
)) {
379 ERR("Health socket poll error");
382 ERR("Unexpected poll events %u for sock %d", revents
, pollfd
);
388 ret
= asprintf(&accepted_socket_name
, "Socket accepted from unix socket @ %s",
389 health_unix_sock_path
);
391 PERROR("Failed to allocate name of accepted socket from unix socket @ %s",
392 health_unix_sock_path
);
395 ret
= fd_tracker_open_unsuspendable_fd(the_fd_tracker
, &new_sock
,
396 (const char **) &accepted_socket_name
, 1,
397 accept_unix_socket
, &sock
);
398 free(accepted_socket_name
);
404 * Set the CLOEXEC flag. Return code is useless because either way, the
407 (void) utils_set_fd_cloexec(new_sock
);
409 DBG("Receiving data from client for health...");
410 ret
= lttcomm_recv_unix_sock(new_sock
, (void *)&msg
, sizeof(msg
));
412 DBG("Nothing recv() from client... continuing");
413 ret
= fd_tracker_close_unsuspendable_fd(the_fd_tracker
,
414 &new_sock
, 1, fd_tracker_util_close_fd
,
425 assert(msg
.cmd
== HEALTH_CMD_CHECK
);
427 memset(&reply
, 0, sizeof(reply
));
428 for (i
= 0; i
< NR_HEALTH_RELAYD_TYPES
; i
++) {
430 * health_check_state return 0 if thread is in
433 if (!health_check_state(health_relayd
, i
)) {
434 reply
.ret_code
|= 1ULL << i
;
438 DBG2("Health check return value %" PRIx64
, reply
.ret_code
);
440 ret
= send_unix_sock(new_sock
, (void *) &reply
, sizeof(reply
));
442 ERR("Failed to send health data back to client");
445 /* End of transmission */
446 ret
= fd_tracker_close_unsuspendable_fd(the_fd_tracker
,
447 &new_sock
, 1, fd_tracker_util_close_fd
,
456 lttng_relay_stop_threads();
459 ERR("Health error occurred in %s", __func__
);
461 DBG("Health check thread dying");
462 unlink(health_unix_sock_path
);
471 * We do NOT rmdir rundir nor the relayd path because there are
472 * other processes using them.
475 lttng_poll_clean(&events
);
477 rcu_unregister_thread();