aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
authorKurt Hackel <kurt.hackel@oracle.com>2006-05-01 16:47:50 -0400
committerMark Fasheh <mark.fasheh@oracle.com>2006-06-26 17:43:08 -0400
commitc8df412e1c746dd21094966d04b3a79aad0f4d08 (patch)
treeb662c0c5b9ff4e5a607d53adb92a8c38acce940e /fs
parent36407488b1cbc4d84bc2bd14e03f3f9b768090d9 (diff)
ocfs2: special case recovery lock in dlmlock_remote()
If the previous master of the recovery lock dies, let calc_usage take it down completely and let the caller completely redo the dlmlock() call. Otherwise, there will never be an opportunity to re-master the lockres and recovery wont be able to progress. Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com> Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Diffstat (limited to 'fs')
-rw-r--r--fs/ocfs2/dlm/dlmlock.c33
-rw-r--r--fs/ocfs2/dlm/dlmrecovery.c4
2 files changed, 27 insertions, 10 deletions
diff --git a/fs/ocfs2/dlm/dlmlock.c b/fs/ocfs2/dlm/dlmlock.c
index 0ff934874942..20b38dc18736 100644
--- a/fs/ocfs2/dlm/dlmlock.c
+++ b/fs/ocfs2/dlm/dlmlock.c
@@ -227,7 +227,16 @@ static enum dlm_status dlmlock_remote(struct dlm_ctxt *dlm,
227 res->state &= ~DLM_LOCK_RES_IN_PROGRESS; 227 res->state &= ~DLM_LOCK_RES_IN_PROGRESS;
228 lock->lock_pending = 0; 228 lock->lock_pending = 0;
229 if (status != DLM_NORMAL) { 229 if (status != DLM_NORMAL) {
230 if (status != DLM_NOTQUEUED) { 230 if (status == DLM_RECOVERING &&
231 dlm_is_recovery_lock(res->lockname.name,
232 res->lockname.len)) {
233 /* recovery lock was mastered by dead node.
234 * we need to have calc_usage shoot down this
235 * lockres and completely remaster it. */
236 mlog(0, "%s: recovery lock was owned by "
237 "dead node %u, remaster it now.\n",
238 dlm->name, res->owner);
239 } else if (status != DLM_NOTQUEUED) {
231 /* 240 /*
232 * DO NOT call calc_usage, as this would unhash 241 * DO NOT call calc_usage, as this would unhash
233 * the remote lockres before we ever get to use 242 * the remote lockres before we ever get to use
@@ -691,18 +700,22 @@ retry_lock:
691 msleep(100); 700 msleep(100);
692 /* no waiting for dlm_reco_thread */ 701 /* no waiting for dlm_reco_thread */
693 if (recovery) { 702 if (recovery) {
694 if (status == DLM_RECOVERING) { 703 if (status != DLM_RECOVERING)
695 mlog(0, "%s: got RECOVERING " 704 goto retry_lock;
696 "for $REOCVERY lock, master " 705
697 "was %u\n", dlm->name, 706 mlog(0, "%s: got RECOVERING "
698 res->owner); 707 "for $RECOVERY lock, master "
699 dlm_wait_for_node_death(dlm, res->owner, 708 "was %u\n", dlm->name,
700 DLM_NODE_DEATH_WAIT_MAX); 709 res->owner);
701 } 710 /* wait to see the node go down, then
711 * drop down and allow the lockres to
712 * get cleaned up. need to remaster. */
713 dlm_wait_for_node_death(dlm, res->owner,
714 DLM_NODE_DEATH_WAIT_MAX);
702 } else { 715 } else {
703 dlm_wait_for_recovery(dlm); 716 dlm_wait_for_recovery(dlm);
717 goto retry_lock;
704 } 718 }
705 goto retry_lock;
706 } 719 }
707 720
708 if (status != DLM_NORMAL) { 721 if (status != DLM_NORMAL) {
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index 86199f66eb56..00209f4a2916 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -2314,6 +2314,10 @@ again:
2314 mlog(0, "%s: reco master %u is ready to recover %u\n", 2314 mlog(0, "%s: reco master %u is ready to recover %u\n",
2315 dlm->name, dlm->reco.new_master, dlm->reco.dead_node); 2315 dlm->name, dlm->reco.new_master, dlm->reco.dead_node);
2316 status = -EEXIST; 2316 status = -EEXIST;
2317 } else if (ret == DLM_RECOVERING) {
2318 mlog(0, "dlm=%s dlmlock says master node died (this=%u)\n",
2319 dlm->name, dlm->node_num);
2320 goto again;
2317 } else { 2321 } else {
2318 struct dlm_lock_resource *res; 2322 struct dlm_lock_resource *res;
2319 2323