Cluster Project branch, master, updated. cluster-2.99.00-43-g8226258
teigland@sourceware.org
teigland@sourceware.org
Tue May 13 21:37:00 GMT 2008
This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "Cluster Project".
http://sources.redhat.com/git/gitweb.cgi?p=cluster.git;a=commitdiff;h=8226258ee9cbfcb6926c01bfa83c7b902e521b61
The branch, master has been updated
via 8226258ee9cbfcb6926c01bfa83c7b902e521b61 (commit)
from fea674786ea0e4ea900b541c3d5a74b69aa0b5e9 (commit)
Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.
- Log -----------------------------------------------------------------
commit 8226258ee9cbfcb6926c01bfa83c7b902e521b61
Author: David Teigland <teigland@redhat.com>
Date: Tue May 13 16:36:19 2008 -0500
dlm_controld: use started_count to detect remerges
Count then number of times the dlm has been started locally and
include the count in the start messages sent to begin each change.
I think this should work to detect remerges after transient
partitions. For now it just logs a warning; after observing how
it works we can use it to ignore remerged nodes.
Signed-off-by: David Teigland <teigland@redhat.com>
-----------------------------------------------------------------------
Summary of changes:
group/dlm_controld/cpg.c | 113 ++++++++++++++++++--------------------
group/dlm_controld/dlm_daemon.h | 1 +
2 files changed, 55 insertions(+), 59 deletions(-)
diff --git a/group/dlm_controld/cpg.c b/group/dlm_controld/cpg.c
index 38babd6..cfa1f14 100644
--- a/group/dlm_controld/cpg.c
+++ b/group/dlm_controld/cpg.c
@@ -629,6 +629,10 @@ static void cleanup_changes(struct lockspace *ls)
free_cg(ls->started_change);
ls->started_change = cg;
+ ls->started_count++;
+ if (!ls->started_count)
+ ls->started_count++;
+
cg->combined_seq = cg->seq; /* for queries */
list_for_each_entry_safe(cg, safe, &ls->changes, list) {
@@ -686,24 +690,30 @@ static void set_plock_ckpt_node(struct lockspace *ls)
ls->plock_ckpt_node = low;
}
+#define COUNT_NUMS 5
+
/* do the change details in the message match the details of the given change */
static int match_change(struct lockspace *ls, struct change *cg,
- struct dlm_header *hd, int len)
+ struct dlm_header *hd, int len, uint32_t *started)
{
struct member *memb;
int member_count, joined_count, remove_count, failed_count;
int i, n_ints, *nums, nodeid, members_mismatch;
uint32_t seq = hd->msgdata;
+ uint32_t started_count;
nums = (int *)((char *)hd + sizeof(struct dlm_header));
- member_count = le32_to_cpu(nums[0]);
- joined_count = le32_to_cpu(nums[1]);
- remove_count = le32_to_cpu(nums[2]);
- failed_count = le32_to_cpu(nums[3]);
+ started_count = le32_to_cpu(nums[0]);
+ member_count = le32_to_cpu(nums[1]);
+ joined_count = le32_to_cpu(nums[2]);
+ remove_count = le32_to_cpu(nums[3]);
+ failed_count = le32_to_cpu(nums[4]);
+
+ *started = started_count;
- n_ints = 4 + member_count;
+ n_ints = COUNT_NUMS + member_count;
if (len < (sizeof(struct dlm_header) + (n_ints * sizeof(int)))) {
log_group(ls, "match_change fail %d:%u bad len %d nums %s",
hd->nodeid, seq, len, str_nums(nums, n_ints));
@@ -715,7 +725,7 @@ static int match_change(struct lockspace *ls, struct change *cg,
get messages for changes prior to the change in which we're added. */
for (i = 0; i < member_count; i++) {
- if (our_nodeid == le32_to_cpu(nums[4+i]))
+ if (our_nodeid == le32_to_cpu(nums[COUNT_NUMS + i]))
break;
}
if (i == member_count) {
@@ -749,7 +759,7 @@ static int match_change(struct lockspace *ls, struct change *cg,
members_mismatch = 0;
for (i = 0; i < member_count; i++) {
- nodeid = le32_to_cpu(nums[4+i]);
+ nodeid = le32_to_cpu(nums[COUNT_NUMS + i]);
memb = find_memb(cg, nodeid);
if (memb)
continue;
@@ -772,7 +782,7 @@ static void send_plocks_stored(struct lockspace *ls)
int n_ints, len, *p, i;
char *buf;
- n_ints = 4 + cg->member_count;
+ n_ints = COUNT_NUMS + cg->member_count;
len = sizeof(struct dlm_header) + (n_ints * sizeof(uint32_t));
buf = malloc(len);
@@ -791,12 +801,13 @@ static void send_plocks_stored(struct lockspace *ls)
us more certainty in matching stopped messages to the correct
change that they are for */
- p[0] = cpu_to_le32(cg->member_count);
- p[1] = cpu_to_le32(cg->joined_count);
- p[2] = cpu_to_le32(cg->remove_count);
- p[3] = cpu_to_le32(cg->failed_count);
+ p[0] = cpu_to_le32(ls->started_count);
+ p[1] = cpu_to_le32(cg->member_count);
+ p[2] = cpu_to_le32(cg->joined_count);
+ p[3] = cpu_to_le32(cg->remove_count);
+ p[4] = cpu_to_le32(cg->failed_count);
- i = 4;
+ i = COUNT_NUMS;
list_for_each_entry(memb, &cg->members, list)
p[i++] = cpu_to_le32(memb->nodeid);
@@ -808,6 +819,8 @@ static void send_plocks_stored(struct lockspace *ls)
static void receive_plocks_stored(struct lockspace *ls, struct dlm_header *hd,
int len)
{
+ uint32_t started_count;
+
log_group(ls, "receive_plocks_stored %d:%u need_plocks %d",
hd->nodeid, hd->msgdata, ls->need_plocks);
@@ -819,7 +832,7 @@ static void receive_plocks_stored(struct lockspace *ls, struct dlm_header *hd,
the next plocks_stored msg following the current start */
if (!list_empty(&ls->changes) || !ls->started_change ||
- !match_change(ls, ls->started_change, hd, len)) {
+ !match_change(ls, ls->started_change, hd, len, &started_count)) {
log_group(ls, "receive_plocks_stored %d:%u ignore",
hd->nodeid, hd->msgdata);
return;
@@ -851,12 +864,12 @@ static void receive_plocks_stored(struct lockspace *ls, struct dlm_header *hd,
for confchg1 or confchg2? Hopefully by comparing the counts and members. */
static struct change *find_change(struct lockspace *ls, struct dlm_header *hd,
- int len)
+ int len, uint32_t *started_count)
{
struct change *cg;
list_for_each_entry_reverse(cg, &ls->changes, list) {
- if (!match_change(ls, cg, hd, len))
+ if (!match_change(ls, cg, hd, len, started_count))
continue;
return cg;
}
@@ -878,30 +891,18 @@ static int is_added(struct lockspace *ls, int nodeid)
return 0;
}
-/* We require new members (memb->added) to be joining the lockspace
- (memb->joining). New members that are not joining the lockspace can happen
- when the cpg partitions and is then merged back together (shouldn't happen
- in general, but is possible). We label these new members that are not
- joining as "disallowed", and ignore their start message. */
-
-/* The added/joining checks to detect disallowed nodes need to consider
- the whole set of changes that are being started. */
-
-/* Handle spurious joins by ignoring this start message if the node says it's
- not joining (i.e. it's already a member), but we see it being added (i.e.
- it's not already a member) */
-
static void receive_start(struct lockspace *ls, struct dlm_header *hd, int len)
{
struct change *cg;
struct member *memb;
- int joining, added;
uint32_t seq = hd->msgdata;
+ uint32_t started_count;
+ int added;
log_group(ls, "receive_start %d:%u flags %x len %d", hd->nodeid, seq,
hd->flags, len);
- cg = find_change(ls, hd, len);
+ cg = find_change(ls, hd, len, &started_count);
if (!cg)
return;
@@ -914,32 +915,25 @@ static void receive_start(struct lockspace *ls, struct dlm_header *hd, int len)
memb->start_flags = hd->flags;
-#if 0
- joining = (memb->start_flags & DLM_MFLG_JOINING) ? 1 : 0;
added = is_added(ls, hd->nodeid);
- if ((added && !joining) || (!added && joining)) {
- log_error("receive_start %d:%u disallowed added %d joining %d",
- hd->nodeid, seq, added, joining);
+ if (added && started_count) {
+ log_error("receive_start %d:%u add node with started_count %u",
+ hd->nodeid, seq, started_count);
+
+ /* observe this scheme working before using it; I'm not sure
+ that a joining node won't ever see an existing node as added
+ under normal circumstances */
+ /*
memb->disallowed = 1;
return;
+ */
}
-#endif
+
node_history_start(ls, hd->nodeid);
memb->start = 1;
}
-static int we_joined(struct lockspace *ls)
-{
- struct change *cg;
-
- list_for_each_entry(cg, &ls->changes, list) {
- if (cg->we_joined)
- return 1;
- }
- return 0;
-}
-
static void send_start(struct lockspace *ls)
{
struct change *cg = list_first_entry(&ls->changes, struct change, list);
@@ -948,7 +942,7 @@ static void send_start(struct lockspace *ls)
int n_ints, len, *p, i;
char *buf;
- n_ints = 4 + cg->member_count;
+ n_ints = COUNT_NUMS + cg->member_count;
len = sizeof(struct dlm_header) + (n_ints * sizeof(int));
buf = malloc(len);
@@ -961,7 +955,7 @@ static void send_start(struct lockspace *ls)
hd->type = DLM_MSG_START;
hd->msgdata = cg->seq;
- if (we_joined(ls))
+ if (ls->joining)
hd->flags |= DLM_MFLG_JOINING;
if (!ls->need_plocks)
@@ -973,18 +967,19 @@ static void send_start(struct lockspace *ls)
us more certainty in matching stopped messages to the correct
change that they are for */
- p[0] = cpu_to_le32(cg->member_count);
- p[1] = cpu_to_le32(cg->joined_count);
- p[2] = cpu_to_le32(cg->remove_count);
- p[3] = cpu_to_le32(cg->failed_count);
+ p[0] = cpu_to_le32(ls->started_count);
+ p[1] = cpu_to_le32(cg->member_count);
+ p[2] = cpu_to_le32(cg->joined_count);
+ p[3] = cpu_to_le32(cg->remove_count);
+ p[4] = cpu_to_le32(cg->failed_count);
- i = 4;
+ i = COUNT_NUMS;
list_for_each_entry(memb, &cg->members, list)
p[i++] = cpu_to_le32(memb->nodeid);
- log_group(ls, "send_start %u flags %x counts %d %d %d %d", cg->seq,
- hd->flags, cg->member_count, cg->joined_count,
- cg->remove_count, cg->failed_count);
+ log_group(ls, "send_start %u flags %x counts %u %d %d %d %d",
+ cg->seq, hd->flags, ls->started_count, cg->member_count,
+ cg->joined_count, cg->remove_count, cg->failed_count);
dlm_send_message(ls, buf, len);
diff --git a/group/dlm_controld/dlm_daemon.h b/group/dlm_controld/dlm_daemon.h
index b35f655..eee0015 100644
--- a/group/dlm_controld/dlm_daemon.h
+++ b/group/dlm_controld/dlm_daemon.h
@@ -166,6 +166,7 @@ struct lockspace {
int kernel_stopped;
int fs_registered;
uint32_t change_seq;
+ uint32_t started_count;
struct change *started_change;
struct list_head changes;
struct list_head node_history;
hooks/post-receive
--
Cluster Project
More information about the Cluster-cvs
mailing list