Logging: Fix breakage when freeing the log ringbuffer

Commit a82d3ec12f0fda16d6bfa8442a07595de897c10e broke fetching from
the log ringbuffer.  The solution there is still generally good: there
is no need to keep the ringbuffer in children created by
ctdb_fork()... except for those special children that are created to
fetch data from the ringbuffer!

Introduce a new function ctdb_fork_no_free_ringbuffer() that does
everything ctdb_fork() needs to do except free the ringbuffer (i.e. it
is the old ctdb_fork() function).  The new ctdb_fork() function just
calls that function and then frees the ringbuffer in the child.

This means all callers of ctdb_fork() have the convenience of having
the ringbuffer freed.  There are 3 special cases:

* Forking the recovery daemon.  We want to be able to fetch from the
  ringbuffer there.

* The ringbuffer fetching code.  Change the 2 calls in this code (main
  daemon, recovery daemon) to call ctdb_fork_no_free_ringbuffer()
  instead.

While we're here, clear the log ringbuffer when the recovery deamon is
forked, since it will contain a copy of the messages from the main
daemon.

Note to self: always test... even the most obvious patches...  ;-)

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

(This used to be ctdb commit 00db5fa00474f8a83f1aa3b603fd756cc9b49ff4)

ctdb/common/ctdb_fork.c

@@ -27,7 +27,7 @@
  * This function forks a child process and drops the realtime 
  * scheduler for the child process.
  */
-pid_t ctdb_fork(struct ctdb_context *ctdb)
+pid_t ctdb_fork_no_free_ringbuffer(struct ctdb_context *ctdb)
 {
 	pid_t pid;
 	char *process;
@@ -60,8 +60,6 @@ pid_t ctdb_fork(struct ctdb_context *ctdb)
 		}
 		ctdb->can_send_controls = false;
 
-		ctdb_log_ringbuffer_free();
-
 		return 0;
 	}
 
@@ -75,6 +73,17 @@ pid_t ctdb_fork(struct ctdb_context *ctdb)
 	return pid;
 }
 
+pid_t ctdb_fork(struct ctdb_context *ctdb)
+{
+	pid_t pid;
+
+	pid = ctdb_fork_no_free_ringbuffer(ctdb);
+	if (pid == 0) {
+		ctdb_log_ringbuffer_free();
+	}
+
+	return pid;
+}
 
 
 static void ctdb_sigchld_handler(struct tevent_context *ev,

ctdb/common/ctdb_logging.c

@@ -163,7 +163,7 @@ int32_t ctdb_control_get_log(struct ctdb_context *ctdb, TDB_DATA addr)
 	/* spawn a child process to marshall the huge log blob and send it back
 	   to the ctdb tool using a MESSAGE
 	*/
-	child = ctdb_fork(ctdb);
+	child = ctdb_fork_no_free_ringbuffer(ctdb);
 	if (child == (pid_t)-1) {
 		DEBUG(DEBUG_ERR,("Failed to fork a log collector child\n"));
 		return -1;

ctdb/include/ctdb_private.h

@@ -1072,6 +1072,7 @@ void ctdb_restore_scheduler(struct ctdb_context *ctdb);
 
 struct tevent_signal *ctdb_init_sigchld(struct ctdb_context *ctdb);
 pid_t ctdb_fork(struct ctdb_context *ctdb);
+pid_t ctdb_fork_no_free_ringbuffer(struct ctdb_context *ctdb);
 int ctdb_kill(struct ctdb_context *ctdb, pid_t pid, int signum);
 
 int32_t ctdb_control_takeover_ip(struct ctdb_context *ctdb, 

ctdb/server/ctdb_recoverd.c

@@ -2126,7 +2126,7 @@ static void getlog_handler(struct ctdb_context *ctdb, uint64_t srvid,
 	}
 	log_addr = (struct ctdb_get_log_addr *)data.dptr;
 
-	child = ctdb_fork(ctdb);
+	child = ctdb_fork_no_free_ringbuffer(ctdb);
 	if (child == (pid_t)-1) {
 		DEBUG(DEBUG_ERR,("Failed to fork a log collector child\n"));
 		return;
@@ -4014,7 +4014,7 @@ int ctdb_start_recoverd(struct ctdb_context *ctdb)
 
 	ctdb->ctdbd_pid = getpid();
 
-	ctdb->recoverd_pid = ctdb_fork(ctdb);
+	ctdb->recoverd_pid = ctdb_fork_no_free_ringbuffer(ctdb);
 	if (ctdb->recoverd_pid == -1) {
 		return -1;
 	}
@@ -4035,6 +4035,9 @@ int ctdb_start_recoverd(struct ctdb_context *ctdb)
 
 	srandom(getpid() ^ time(NULL));
 
+	/* Clear the log ringbuffer */
+	ctdb_clear_log(ctdb);
+
 	if (switch_from_server_to_client(ctdb, "recoverd") != 0) {
 		DEBUG(DEBUG_CRIT, (__location__ "ERROR: failed to switch recovery daemon into client mode. shutting down.\n"));
 		exit(1);