ocfs2/cluster/quorum.c

12  * that a node is broken and should be recovered.  They can't both recover each
17  * So we declare that a node which has given up on connecting to a majority
20  * There are huge opportunities for races here.  After we give up on a node's
22  * to declare the node as truly dead.  We also need to be careful with the
23  * race between when we see a node start heartbeating and when we connect
83  * go away as our node would be fenced externally before other nodes
115 			mlog(ML_ERROR, "fencing this node because it is "
126 		 * the lowest numbered node */
129 			mlog(ML_ERROR, "fencing this node because it is "
138 			mlog(ML_ERROR, "fencing this node because it is "
141 			     "node %u\n", quorum, qs->qs_heartbeating,
152 		mlog(ML_NOTICE, "not fencing this node, heartbeating: %d, "
162 static void o2quo_set_hold(struct o2quo_state *qs, u8 node)
166 	if (!test_and_set_bit(node, qs->qs_hold_bm)) {
169 			        "node %u\n", node);
170 		mlog(0, "node %u, %d total\n", node, qs->qs_holds);
174 static void o2quo_clear_hold(struct o2quo_state *qs, u8 node)
178 	if (test_and_clear_bit(node, qs->qs_hold_bm)) {
179 		mlog(0, "node %u, %d total\n", node, qs->qs_holds - 1);
186 		mlog_bug_on_msg(qs->qs_holds < 0, "node %u, holds %d\n",
187 				node, qs->qs_holds);
191 /* as a node comes up we delay the quorum decision until we know the fate of
195 void o2quo_hb_up(u8 node)
203 		        "node %u\n", node);
204 	mlog_bug_on_msg(test_bit(node, qs->qs_hb_bm), "node %u\n", node);
205 	set_bit(node, qs->qs_hb_bm);
207 	mlog(0, "node %u, %d total\n", node, qs->qs_heartbeating);
209 	if (!test_bit(node, qs->qs_conn_bm))
210 		o2quo_set_hold(qs, node);
212 		o2quo_clear_hold(qs, node);
217 /* hb going down releases any holds we might have had due to this node from
219 void o2quo_hb_down(u8 node)
227 			"node %u, %d heartbeating\n",
228 			node, qs->qs_heartbeating);
229 	mlog_bug_on_msg(!test_bit(node, qs->qs_hb_bm), "node %u\n", node);
230 	clear_bit(node, qs->qs_hb_bm);
232 	mlog(0, "node %u, %d total\n", node, qs->qs_heartbeating);
234 	o2quo_clear_hold(qs, node);
239 /* this tells us that we've decided that the node is still heartbeating
244 void o2quo_hb_still_up(u8 node)
250 	mlog(0, "node %u\n", node);
253 	o2quo_clear_hold(qs, node);
258 /* This is analogous to hb_up.  as a node's connection comes up we delay the
263 void o2quo_conn_up(u8 node)
271 		        "node %u\n", node);
272 	mlog_bug_on_msg(test_bit(node, qs->qs_conn_bm), "node %u\n", node);
273 	set_bit(node, qs->qs_conn_bm);
275 	mlog(0, "node %u, %d total\n", node, qs->qs_connected);
277 	if (!test_bit(node, qs->qs_hb_bm))
278 		o2quo_set_hold(qs, node);
280 		o2quo_clear_hold(qs, node);
285 /* we've decided that we won't ever be connecting to the node again.  if it's
287  * node stops heartbeating from hb_down or the caller decides that the node is
289 void o2quo_conn_err(u8 node)
295 	if (test_bit(node, qs->qs_conn_bm)) {
298 				"node %u, connected %d\n",
299 				node, qs->qs_connected);
301 		clear_bit(node, qs->qs_conn_bm);
303 		if (test_bit(node, qs->qs_hb_bm))
304 			o2quo_set_hold(qs, node);
307 	mlog(0, "node %u, %d total\n", node, qs->qs_connected);