From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sunil Mushran Date: Mon, 21 Feb 2011 14:25:24 -0800 Subject: [Ocfs2-devel] [PATCH 1/2] ocfs2/cluster: Heartbeat mismatch message improved Message-ID: <1298327125-4045-1-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 If o2hb finds unexpected values in the heartbeat slot, it prints a message "ERROR: Device "dm-6": another node is heartbeating in our slot!" This patch adds more information allowing us to see the actual mismatch. The new message reads "ERROR: Heartbeat mismatch on "dm-6": expected(92:0x234543, 0x76567) ondisk(92:0x234543, 0x76565)" Signed-off-by: Sunil Mushran --- fs/ocfs2/cluster/heartbeat.c | 34 +++++++++++++++++----------------- 1 files changed, 17 insertions(+), 17 deletions(-) diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c index b108e86..d633df0 100644 --- a/fs/ocfs2/cluster/heartbeat.c +++ b/fs/ocfs2/cluster/heartbeat.c @@ -543,25 +543,27 @@ static int o2hb_verify_crc(struct o2hb_region *reg, /* We want to make sure that nobody is heartbeating on top of us -- * this will help detect an invalid configuration. */ -static int o2hb_check_last_timestamp(struct o2hb_region *reg) +static void o2hb_check_last_timestamp(struct o2hb_region *reg) { - int node_num, ret; struct o2hb_disk_slot *slot; struct o2hb_disk_heartbeat_block *hb_block; - node_num = o2nm_this_node(); - - ret = 1; - slot = ®->hr_slots[node_num]; + slot = ®->hr_slots[o2nm_this_node()]; /* Don't check on our 1st timestamp */ - if (slot->ds_last_time) { - hb_block = slot->ds_raw_block; + if (!slot->ds_last_time) + return; - if (le64_to_cpu(hb_block->hb_seq) != slot->ds_last_time) - ret = 0; - } + hb_block = slot->ds_raw_block; + if (le64_to_cpu(hb_block->hb_seq) == slot->ds_last_time) + return; - return ret; + mlog(ML_ERROR, "Heartbeat mismatch on \"%s\": " + "expected(%u:0x%llx, 0x%llx), ondisk(%u:0x%llx, 0x%llx)\n", + reg->hr_dev_name, slot->ds_node_num, + (unsigned long long)slot->ds_last_generation, + (unsigned long long)slot->ds_last_time, hb_block->hb_node, + (unsigned long long)le64_to_cpu(hb_block->hb_generation), + (unsigned long long)le64_to_cpu(hb_block->hb_seq)); } static inline void o2hb_prepare_block(struct o2hb_region *reg, @@ -987,9 +989,7 @@ static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) /* With an up to date view of the slots, we can check that no * other node has been improperly configured to heartbeat in * our slot. */ - if (!o2hb_check_last_timestamp(reg)) - mlog(ML_ERROR, "Device \"%s\": another node is heartbeating " - "in our slot!\n", reg->hr_dev_name); + o2hb_check_last_timestamp(reg); /* fill in the proper info for our next heartbeat */ o2hb_prepare_block(reg, reg->hr_generation); @@ -1003,8 +1003,8 @@ static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) } i = -1; - while((i = find_next_bit(configured_nodes, O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) { - + while((i = find_next_bit(configured_nodes, + O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) { change |= o2hb_check_slot(reg, ®->hr_slots[i]); } -- 1.7.1