From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sunil Mushran Date: Tue, 14 Sep 2010 15:50:46 -0700 Subject: [Ocfs2-devel] [PATCH 10/20] ocfs2/cluster: Check slots for unconfigured live nodes In-Reply-To: <1284504656-2434-1-git-send-email-sunil.mushran@oracle.com> References: <1284504656-2434-1-git-send-email-sunil.mushran@oracle.com> Message-ID: <1284504656-2434-11-git-send-email-sunil.mushran@oracle.com> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ocfs2-devel@oss.oracle.com o2hb currently checks slots for configured nodes only. This patch makes it check the slots for the live nodes too to take care of a race in which a node is removed from the configuration but not from the live map. Signed-off-by: Sunil Mushran --- fs/ocfs2/cluster/heartbeat.c | 38 +++++++++++++++++++++++++++++++------- 1 files changed, 31 insertions(+), 7 deletions(-) diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c index 1d71856..de798c7 100644 --- a/fs/ocfs2/cluster/heartbeat.c +++ b/fs/ocfs2/cluster/heartbeat.c @@ -593,14 +593,24 @@ static int o2hb_check_slot(struct o2hb_region *reg, u64 cputime; unsigned int dead_ms = o2hb_dead_threshold * O2HB_REGION_TIMEOUT_MS; unsigned int slot_dead_ms; + int tmp; memcpy(hb_block, slot->ds_raw_block, reg->hr_block_bytes); - /* Is this correct? Do we assume that the node doesn't exist - * if we're not configured for him? */ + /* + * If a node is no longer configured but is still in the livemap, we + * may need to clear that bit from the livemap. + */ node = o2nm_get_node_by_num(slot->ds_node_num); - if (!node) - return 0; + if (!node) { + spin_lock(&o2hb_live_lock); + tmp = test_bit(slot->ds_node_num, o2hb_live_node_bitmap); + spin_unlock(&o2hb_live_lock); + if (!tmp) + return 0; + printk(KERN_NOTICE "o2hb: Live node %d is not registered\n", + slot->ds_node_num); + } if (!o2hb_verify_crc(reg, hb_block)) { /* all paths from here will drop o2hb_live_lock for @@ -717,8 +727,9 @@ fire_callbacks: if (list_empty(&o2hb_live_slots[slot->ds_node_num])) { clear_bit(slot->ds_node_num, o2hb_live_node_bitmap); - o2hb_queue_node_event(&event, O2HB_NODE_DOWN_CB, node, - slot->ds_node_num); + if (node) + o2hb_queue_node_event(&event, O2HB_NODE_DOWN_CB, + node, slot->ds_node_num); changed = 1; } @@ -738,7 +749,8 @@ out: o2hb_run_event_list(&event); - o2nm_node_put(node); + if (node) + o2nm_node_put(node); return changed; } @@ -765,6 +777,7 @@ static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) { int i, ret, highest_node, change = 0; unsigned long configured_nodes[BITS_TO_LONGS(O2NM_MAX_NODES)]; + unsigned long live_node_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; struct o2hb_bio_wait_ctxt write_wc; ret = o2nm_configured_node_map(configured_nodes, @@ -774,6 +787,17 @@ static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) return ret; } + /* + * If a node is not configured but is in the livemap, we still need + * to read the slot so as to be able to remove it from the livemap. + */ + o2hb_fill_node_map(live_node_bitmap, sizeof(live_node_bitmap)); + i = -1; + while((i = find_next_bit(live_node_bitmap, + O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) { + set_bit(i, configured_nodes); + } + highest_node = o2hb_highest_node(configured_nodes, O2NM_MAX_NODES); if (highest_node >= O2NM_MAX_NODES) { mlog(ML_NOTICE, "ocfs2_heartbeat: no configured nodes found!\n"); -- 1.7.0.4