[RFC PATCH] lvmlockd: purge the lock resources left in previous lockspace

* [RFC PATCH] lvmlockd: purge the lock resources left in previous lockspace
@ 2022-09-30 13:57 Lidong Zhong
  2022-09-30 15:32 ` David Teigland
  0 siblings, 1 reply; 3+ messages in thread
From: Lidong Zhong @ 2022-09-30 13:57 UTC (permalink / raw)
  To: lvm-devel

If lvmlockd in cluster is killed accidently or any other reason, the
lock resources will become orphaned in the VG lockspace. When the
cluster manager tries to restart this daemon, the LVs will probably
become inactive because of resource schedule policy and thus the lock
resouce will be omited during the adoption process. This patch will
try to purge the lock resources left in previous lockspace, so the
following actions can work again.
---
 ?daemons/lvmlockd/lvmlockd-core.c???? | 12 +++++
 ?daemons/lvmlockd/lvmlockd-dlm.c????? | 77 ++++++++++++++++++++++++++++
 ?daemons/lvmlockd/lvmlockd-internal.h |? 6 +++
 ?3 files changed, 95 insertions(+)

diff --git a/daemons/lvmlockd/lvmlockd-core.c 
b/daemons/lvmlockd/lvmlockd-core.c
index 6d0d4d98c..054bd9a10 100644
--- a/daemons/lvmlockd/lvmlockd-core.c
+++ b/daemons/lvmlockd/lvmlockd-core.c
@@ -5955,6 +5955,18 @@ static void adopt_locks(void)


 ???? /* FIXME: purge any remaining orphan locks in each rejoined ls? */
+??? /* Try to purge the orphan locks when lock manager is dlm */
+??? if (lm_support_dlm() && lm_is_running_dlm()) {
+??? ??? list_for_each_entry(ls, &ls_found, list) {
+??? ??? ??? pthread_mutex_lock(&lockspaces_mutex);
+??? ??? ??? ls1 = find_lockspace_name(ls->name);
+??? ??? ??? if (ls1) {
+??? ??? ??? ??? log_debug("ls: %s purge locks", ls->name);
+??? ??? ??? ??? lm_purge_locks_dlm(ls1);
+??? ??? ??? }
+??? ??? ??? pthread_mutex_unlock(&lockspaces_mutex);
+??? ??? }
+??? }

 ???? if (count_start_fail || count_adopt_fail)
 ???? ??? goto fail;
diff --git a/daemons/lvmlockd/lvmlockd-dlm.c 
b/daemons/lvmlockd/lvmlockd-dlm.c
index 1305c3dc2..f63f5ec82 100644
--- a/daemons/lvmlockd/lvmlockd-dlm.c
+++ b/daemons/lvmlockd/lvmlockd-dlm.c
@@ -220,6 +220,83 @@ int lm_prepare_lockspace_dlm(struct lockspace *ls)
 ???? return 0;
 ?}

+#define DLM_COMMS_PATH "/sys/kernel/config/dlm/cluster/comms"
+#define LOCK_LINE_MAX 1024
+static int get_local_nodeid()
+{
+??? struct dirent *de;
+??? DIR *ls_dir;
+??? char ls_comms_path[PATH_MAX];
+??? FILE *file = NULL;
+??? char line[LOCK_LINE_MAX];
+??? int rv = -1, val;
+
+??? memset(ls_comms_path, 0, sizeof(ls_comms_path));
+??? snprintf(ls_comms_path, PATH_MAX, "%s",DLM_COMMS_PATH);
+
+??? if (!(ls_dir = opendir(ls_comms_path)))
+??? ??? return -ECONNREFUSED;
+
+??? while ((de = readdir(ls_dir))) {
+??? ??? if (de->d_name[0] == '.')
+??? ??? ??? continue;
+??? ??? memset(ls_comms_path, 0, sizeof(ls_comms_path));
+??? ??? snprintf(ls_comms_path, PATH_MAX, "%s/%s/local",
+??? ??? ???? DLM_COMMS_PATH, de->d_name);
+??? ??? file = fopen(ls_comms_path, "r");
+??? ??? if (!file)
+??? ??? ??? continue;
+??? ??? if (fgets(line, LOCK_LINE_MAX, file)) {
+??? ??? ??? fclose(file);
+??? ??? ??? rv = sscanf(line, "%d", &val);
+??? ??? ??? if ((rv == 1) && (val == 1 )) {
+??? ??? ??? ??? memset(ls_comms_path, 0, sizeof(ls_comms_path));
+??? ??? ??? ??? snprintf(ls_comms_path, PATH_MAX, "%s/%s/nodeid",
+??? ??? ??? ??? ??? DLM_COMMS_PATH, de->d_name);
+??? ??? ??? ??? file = fopen(ls_comms_path, "r");
+??? ??? ??? ??? if (!file)
+??? ??? ??? ??? ??? continue;
+??? ??? ??? ??? if (fgets(line, LOCK_LINE_MAX, file)) {
+??? ??? ??? ??? ??? rv = sscanf(line, "%d", &val);
+??? ??? ??? ??? ??? if (rv == 1)
+??? ??? ??? ??? ??? ??? return val;
+??? ??? ??? ??? }
+??? ??? ??? }
+??? ??? }
+??? ??? fclose(file);
+??? }
+
+??? if (closedir(ls_dir))
+??? ??? log_error("get_local_nodeid closedir error");
+??? return rv;
+}
+
+int lm_purge_locks_dlm(struct lockspace *ls)
+{
+??? struct lm_dlm *lmd = (struct lm_dlm *)ls->lm_data;
+??? int nodeid;
+??? int rv = -1;
+
+??? if (!lmd || !lmd->dh) {
+??? ??? log_error("purge_locks_dlm %s no dlm_handle_t error", ls->name);
+??? ??? goto fail;
+??? }
+
+??? nodeid = get_local_nodeid();
+??? if (nodeid < 0) {
+??? ??? log_error("failed to get local nodeid");
+??? ??? goto fail;
+??? }
+??? if (dlm_ls_purge(lmd->dh, nodeid, 0)) {
+??? ??? log_error("purge_locks_dlm %s error", ls->name);
+??? ??? goto fail;
+??? }
+
+??? rv = 0;
+fail:
+??? return rv;
+}
+
 ?int lm_add_lockspace_dlm(struct lockspace *ls, int adopt)
 ?{
 ???? struct lm_dlm *lmd = (struct lm_dlm *)ls->lm_data;
diff --git a/daemons/lvmlockd/lvmlockd-internal.h 
b/daemons/lvmlockd/lvmlockd-internal.h
index ad32eb3a4..dd59b6a5d 100644
--- a/daemons/lvmlockd/lvmlockd-internal.h
+++ b/daemons/lvmlockd/lvmlockd-internal.h
@@ -392,6 +392,7 @@ static inline const char *mode_str(int x)
 ?int lm_init_vg_dlm(char *ls_name, char *vg_name, uint32_t flags, char 
*vg_args);
 ?int lm_prepare_lockspace_dlm(struct lockspace *ls);
 ?int lm_add_lockspace_dlm(struct lockspace *ls, int adopt);
+int lm_purge_locks_dlm(struct lockspace *ls);
 ?int lm_rem_lockspace_dlm(struct lockspace *ls, int free_vg);
 ?int lm_lock_dlm(struct lockspace *ls, struct resource *r, int ld_mode,
 ???? ??? struct val_blk *vb_out, int adopt);
@@ -429,6 +430,11 @@ static inline int lm_add_lockspace_dlm(struct 
lockspace *ls, int adopt)
 ???? return -1;
 ?}

+static inline int lm_purge_locks_dlm(struct lockspace *ls)
+{
+??? return -1;
+}
+
 ?static inline int lm_rem_lockspace_dlm(struct lockspace *ls, int free_vg)
 ?{
 ???? return -1;
-- 
2.35.3


^ permalink raw reply related	[flat|nested] 3+ messages in thread